Nos. 24-7433 25-2204

Agency No. A092-078-794

On Petition for Review of an Order of the Board of Immigration Appeals

Argued and Submitted May 19, 2026 San Francisco, California

Filed August 28, 2026

Before: Kim McLane Wardlaw, Carlos T. Bea, and Gabriel P. Sanchez, Circuit Judges.

Opinion by Judge Wardlaw; Dissent by Judge Bea

Page 2SUMMARY *

Granting Alfredo Parada Calderon’s petition for review of a decision of the Board of Immigration Appeals that denied his motion to reopen removal proceedings to submit additional evidence in support of his claim for protection under the Convention Against Torture (“CAT”), the panel held that the BIA erred by applying a “quantitative evidence' requirement and by failing to give reasoned consideration to Parada’s expert reports.

The BIA found that Parada demonstrated a reasonable likelihood that he could be detained in El Salvador, as a deportee, under the crackdown on civil liberties known as the “State of Exception.” However, the BIA dismissed some of Parada’s evidence, including an expert report’s descriptions of “deplorable conditions of men in prison,” on the basis that “anecdotes—no matter how disturbing—can’t substitute for quantitative evidence, which is what CAT requires.” The panel wrote that Amici before this court had documented a growing and troubling trend in CAT adjudications: the demand for statistical or quantitative proof to establish that an applicant is more likely than not to be tortured.

The panel concluded that the BIA’s imposition of this “quantitative evidence” requirement demonstrated a fundamental misunderstanding of CAT's implementing regulations, which require the BIA to consider “all evidence Page 3relevant to the possibility of future torture.'⁸ C.F.R. § 1208.16(c)(3). The BIA’s disregard of the unambiguous language of the regulations imposed a nearly insurmountable burden on Parada by requiring him to produce statistics of torture in El Salvador that likely do not exist or, if any do exist, are likely unreliable.

The panel also concluded that the BIA erred by failing to give reasoned consideration to Parada’s expert reports. The BIA failed to discuss two of the three reports and, as to the one it discussed, the BIA dismissed the report as lacking the allegedly requisite “quantitative evidence.”

The panel concluded that the record evidence compels the conclusion that Parada established a reasonable likelihood of showing, in a reopened proceeding, that he is more likely than not to be tortured if he were removed to El Salvador. The record reflected that officials directly perpetrate and acquiesce to widespread acts of torture and other abuses in Salvadoran prisons, including beatings, starvation, electric shocking, sexual assaults, the denial of medical treatment, and extrajudicial killings. Here, Parada’s individualized evidence-such as his criminal history, association with multiple gangs, and tattoos—was sufficient to establish prima facie eligibility for protection under CAT.

The panel remanded to the BIA with instructions to grant his motion to reopen.

In a concurrently filed memorandum disposition, the panel resolved Parada's petition for review of the BIA's denial of his applications for asylum, withholding of removal, and protection under CAT.

Dissenting, Judge Bea wrote that the BIA did not impose a 'quantitative evidence' requirement. Rather, the BIAPage 4 recited verbatim a passage from this court's decision in Benedicto v. Garland , 12 F.4th 1049 (9th Cir. 2021). Judge Bea wrote that, as a three-judge panel, it had no authority to overrule Benedicto , so it could not be that the BIA abuses its discretion when it correctly adheres to that precedent.

Regarding consideration of Parada's expert reports, Judge Bea wrote that the majority ignored the longstanding practice of presuming that public officers properly discharge their official duties. Judge Bea also wrote that this court— recognizing the realities of immigration proceedings—does not require the BIA to discuss each piece of evidence submitted.

In Judge Bea's view, the majority broke new ground because this court has repeatedly upheld denials of CAT applications based on similar State of Exception evidence and arguments, doing so in unpublished decisions without any noted dissents.

Page 5Genna E. Beier (argued) and Sean L. McMahon, Deputy Public Defenders; Matt Gonzalez, Chief Attorney; Manohar Raju, Public Defender; Office of the San Francisco Public Defender, San Francisco, California; for Petitioner.

Anne R. Burley (argued) and Jennifer P. Williams, Trial Attorneys; Song Park, Assistant Director; Office of Immigration Litigation; Brett A. Shumate, Assistant Attorney General; Civil Division, United States Department of Justice, Washington, D.C.; for Respondent.

Nancy Alexander, Lucas & Barba LLP, Portland, Oregon; Veronica Barba, Lucas & Barba LLP, Pasadena, California; for Amicus Curiae Refugees International.

Ashley B. Vinson Crawford, Akin Gump Strauss Hauer & Feld LLP, San Francisco, California, for Amici Curiae Former Immigration Judges & Former Members of the Board of Immigration Appeals.

Alletta S. Brenner and Sharila Stewart, Ashurst Perkins Coie US LLP, Portland, Oregon, for Amicus Curiae Graeme Blair.

Page 6WARDLAW, Circuit Judge:

This case is about the increasingly widespread misapplication of plain and unambiguous legal standards by our immigration courts. Alfredo Parada Calderon (“Parada”), a native and citizen of El Salvador, petitions for review of two separate decisions of the Board of Immigration Appeals (“BIA”). He first seeks review of the BIA's dismissal of his appeal of an Immigration Judge's (“IJ”) denial of his applications for asylum, withholding of removal, and protection under the Convention Against Torture ('CAT'). Separately, Parada seeks review of the BIA’s denial of his motion to reopen removal proceedings. We have jurisdiction over both petitions under 8 U.S.C. § 1252(a).

The BIA committed two clear legal errors in its denial of Parada’s motion to reopen. First, in line with a growing and troubling trend across the Nation, the BIA held that, to obtain CAT protection, a petitioner must produce 'quantitative evidence” and that, without such evidence, Parada could not meet his burden of showing that he was entitled to CAT protection.¹ Those misstatements of law demonstrate a Page 7fundamental misunderstanding of CAT's implementing regulations, which expressly require the BIA to consider “all evidence relevant to the possibility of future torture' in adjudicating a noncitizen’s CAT application. See 8 C.F.R. § 1208.16(c)(3). The BIA’s disregard of the unambiguous language of the regulations imposed a nearly insurmountable burden on Parada by requiring him to produce statistics of torture in El Salvador that likely do not exist or, if any do exist, are likely unreliable. Second, the BIA erred by failing to give reasoned consideration to Parada's expert reports. See Cole v. Holder , 659 F.3d 762, 772 (9th Cir. 2011). Under the correct legal standards, the record evidence compels the conclusion that Parada met the burden of showing that he has a “reasonable likelihood of [succeeding] on the merits” of his application in a reopened proceeding. See Fonseca Fonseca v. Garland , 76 F.4th 1176, 1181 (9th Cir. 2023) (citation modified). Accordingly, we grant Parada’s petition to review the BIA’s denial of his motion to reopen and remand to the BIA with instructions to grant his motion to reopen.²

Parada was placed in removal proceedings in 2023. An IJ denied his application for CAT protection in February 2024. On appeal, the BIA remanded the record to the IJ for Page 8further proceedings because the IJ had erred by refusing to consider certain country conditions evidence and by failing to adequately explain why she had afforded reduced weight to an expert report. On remand, the IJ denied Parada's request to reopen the record and denied his application for CAT protection. The BIA denied Parada’s appeal of that decision. Parada filed a petition for review of the BIA's decision in our court.

While his petition was pending, on February 24, 2025, Parada filed a motion to reopen his removal proceedings, seeking to submit additional evidence of torture in El Salvador. The motion, though filed two days after the 90day deadline, was deemed timely after the BIA equitably tolled the deadline based on Parada's lawyer's ineffective assistance.

The BIA denied his motion to reopen on the merits. The BIA acknowledged that Parada submitted 'new evidence provid[ing] more recent information on the treatment of deportees upon return to El Salvador,” and found that Parada 'demonstrated a reasonable likelihood that he could be detained in El Salvador under the current state of exception.” 3 However, the BIA dismissed some of the evidence Parada submitted in support of his motion to reopen, including an expert report’s descriptions of “deplorable conditions of men in prison,” on the basis that “anecdotes—no matter how disturbing—can’t substitute for Page 9quantitative evidence, which is what CAT requires.” The BIA rejected 'the majority of the evidence' that Parada submitted on the grounds that this evidence was “not based on new or previously unavailable evidence but reflect[ed] the continuation of abuse of suspected gang members under the state of exception on March 27, 2022, which was prior to [Parada’s] individual hearing.” The BIA concluded, based on the evidence it did consider, that “the evidence does not show a reasonable likelihood that [Parada] would be able to establish that it is more likely than not that he would be tortured if detained,' and denied his motion to reopen. Parada petitioned for review of the BIA's denial of his motion to reopen.

We review the denial of a motion to reopen for abuse of discretion.⁴ Hernandez-Velasquez v. Holder , 611 F.3d 1073, 1077 (9th Cir. 2010). 'An error of law is an abuse of discretion.' Id . Factual findings underlying the adjudication of a motion to reopen are reviewed for substantial evidence, while legal questions are reviewed de novo. Bhasin v. Gonzales , 423 F.3d 977, 983 (9th Cir. 2005). We review the IJ’s denial of protection under CAT for substantial evidence.

Page 10Duran-Rodriguez v. Barr , 918 F.3d 1025, 1028 (9th Cir. 2019).

“[I]t [is] the policy of the United States not to [remove] any person to a country in which there are substantial grounds for believing the person would be in danger of being subjected to torture.” Trinidad y Garcia v. Thomas , 683 F.3d 952, 956 (9th Cir. 2012) (en banc) (per curiam) (citation modified) (citing 8 U.S.C. § 1231 note). Accordingly, the key question in evaluating a CAT claim is 'whether [the noncitizen] has shown individual circumstances in his case that there is a likelihood he will face torture qualifying him for CAT.” Ridore v. Holder , 696 F.3d 907, 912 (9th Cir. 2012) (internal quotation marks and citation omitted). However, the nature of CAT is that it provides only limited protection from removal—nothing prevents the Government from removing a noncitizen “to a third country other than the country to which removal has been withheld or deferred.” See 8 C.F.R. § 1208.16(f). CAT thus strikes a delicate balance: regardless of the strength of the Government's interest in removing a noncitizen, he cannot be removed to a country in which he is more likely than not to face torture with the consent or acquiescence of the government. Id. But CAT protection will not prevent the Government from removing him to a third country in which he is not more likely than not to face torture. Id.

In this case, the BIA erred as a matter of law by requiring Parada to produce “quantitative evidence” of his likelihood of experiencing future torture in El Salvador. Such a requirement is irreconcilable with the nature of CAT protection, the plain text of CAT’s implementing regulations, and the Treaty’s purpose. See 8 C.F.R.

Page 11§ 1208.16(c)(3). CAT requires an individualized inquiry based on all of the evidence in the record. Contrary to the suggestion of the dissent, even if Parada could come forward with “evidence that [more than] half of the detainees in El Salvador are tortured with government consent or acquiescence,” Dissent at 43, that would prove nothing about the individualized circumstances that render him particularly more likely than not to be tortured.

Moreover, the BIA committed the paradigmatic error that we warned against in Cole v. Holder by failing to give reasoned explanation to the three expert reports offered by Parada. 659 F.3d at 771-72. The record compels the conclusion that Parada met his burden to prevail on a motion to reopen because he demonstrated “a reasonable likelihood that the petitioner would prevail on the merits if the motion to reopen were granted.' Fonseca-Fonseca , 76 F.4th at 1179.

The United States “implemented the CAT as part of the Foreign Affairs Reform and Restructuring Act of 1998' (“FARRA”). Sridej v. Blinken , 108 F.4th 1088, 1090 (9th Cir. 2024). Under CAT's implementing regulations, the BIA is required to consider 'all evidence relevant to the possibility of future torture . . . including, but not limited to: (i) [e]vidence of past torture inflicted upon the applicant; (ii) [e]vidence that the applicant could relocate to a part of the country of removal where he or she is not likely to be tortured; (iii) [e]vidence of gross, flagrant or mass violations of human rights within the country of removal, where applicable; and (iv) [o]ther relevant information regarding conditions in the country of removal.” 8 C.F.R.

Page 12§ 1208.16(c)(3); Uc Encarnacion v. Bondi , 156 F.4th 927, 941 (9th Cir. 2025) ('The regulations implementing CAT explicitly require the IJ to consider all evidence relevant to the possibility of future torture.” (citation modified)).

CAT's regulatory mandate is purposefully expansive and rejects a one-size-fits-all approach to establishing eligibility for relief. By requiring the agency to consider “ all evidence relevant to the possibility of future torture,'⁸ C.F.R. § 1208.16(c)(3) (emphasis added), Congress ensured that there is not only one single path to receiving protection under the Convention.⁵ See Wakkary v. Holder , 558 F.3d 1049, 1068 (9th Cir. 2009) (“[T]he regulations that implement the CAT do not specify two routes . . . by which an applicant may establish [a] likelihood [of future torture]. Rather, the CAT regulations cast a wide evidentiary net, providing that 'all evidence relevant to the possibility of future torture shall be considered.'' (quoting 8 C.F.R. § 1208.16(c)(3)) (internal citation omitted)).

Accordingly, we have previously stated that “[a]n applicant may . . . satisfy the burden to receive CAT relief with evidence of country conditions alone.” Uc Encarnacion , 156 F.4th at 941. Expert witness testimony also can provide “highly probative [and] potentially dispositive evidence” that the BIA must consider in evaluating an application for protection under CAT. Cole , 659 F.3d at 772; see also Uc Encarnacion , 156 F.4th at 943 ('If the Board had accorded Petitioner's expert testimony and country-conditions evidence any weight, it could have concluded that” he was eligible for protection under CAT.).

Page 13Similarly, CAT’s implementing regulations explicitly provide that 'a petitioner's credible testimony 'may be sufficient to sustain the burden of proof without corroboration.'' Parada v. Sessions , 902 F.3d 901, 915 (9th Cir. 2018) (quoting 8 C.F.R. § 1208.16(c)(2)).

Having established that the BIA must consider all evidence of an applicant's risk of torture, we turn to the burden itself: “The burden of proof is on the applicant . . . to establish that it is more likely than not that he or she would be tortured if removed to the proposed country of removal.” 8 C.F.R. § 1208.16(c)(2) (emphasis added). We have long interpreted the 'more likely than not' standard to require petitioners to show “a greater than fifty percent chance of torture.” Edu v. Holder , 624 F.3d 1137, 1145 n.16 (9th Cir. 2010); see also Hamoui v. Ashcroft , 389 F.3d 821, 827 (9th Cir. 2004) (A petitioner “has to show only a chance greater than fifty percent that he will be tortured if removed.'). While we have little doubt that this is the correct interpretation of that phrase, the BIA's application of the “more likely than not” standard has strayed from the plain text and purpose of the regulations.

Amici have documented a “growing and troubling trend” in CAT adjudications: the 'demand for statistical or quantitative proof to establish that an applicant is 'more likely than not' to be tortured.' See Amicus Brief of Immigration Judges, at 1. In this case, the BIA rejected significant evidence of torture in El Salvador, explaining that: “While we do not condone the treatment described in the evidence presented, ‘[a]necdotes—no matter how disturbing—can’t substitute for quantitative evidence, which is what CAT requires.’” In imposing a quantitative evidence condition, the BIA conflated two distinct requirements in the regulation: (i) that the BIA must consider all types of Page 14evidence, and (ii) that a petitioner must produce evidence sufficient to satisfy the BIA that it is more likely than not , judged holistically, in light of all evidence in the record, that the individual petitioner will, himself, be tortured if removed.⁸ C.F.R. § 1208.16(c)(2); see also Cole , 659 F.3d at 775 (the BIA must consider the “aggregate risk” of future torture).

The plain text of CAT's implementing regulations confirms that quantitative or statistical evidence is not required. Such a requirement is irreconcilable with the regulatory mandate that “all evidence relevant to the possibility of future torture shall be considered.” 8 C.F.R. § 1208.16(c)(3). Congress did not create a bespoke evidentiary standard whereby noncitizens must come into court armed with reports prepared by statisticians and country conditions experts prognosticating about the noncitizen’s exact numerical likelihood of being tortured.⁶ Congress instructed the agency to perform a much less foreign task: Immigration judges must make 'a reasoned, fact-based judgment that [the petitioner's] torture is more Page 15likely than not, based on the totality of the circumstances.” See Amicus Brief of Immigration Judges, at 7.⁷

CAT's implementing regulations do not elevate the significance or probative weight of statistical evidence above any other form of evidence in the record. Indeed, Page 16amici provide compelling reasons to doubt the reliability of statistical evidence as an illustrative tool to estimate a noncitizen's individualized risk of being tortured. See generally Amicus Brief of Professor Blair; see also Amicus Brief of Refugees International, at 3–8. The problems with relying on statistical evidence are at their most acute in repressive regimes, like El Salvador. See Amicus Brief of Professor Blair, at 15.

As we have long recognized, “[t]here is nothing novel about the concept that persecutors [or torturers] cannot be expected to conform to arbitrary evidentiary rules.” Aguilera-Cota v. U.S. Immigr. & Nat’y Serv. , 914 F.2d 1375, 1380 (9th Cir. 1990). Torturers and '[p]ersecutors are hardly likely to provide their victims with affidavits attesting to their acts of persecution.” Bolanos-Hernandez v. I.N.S. , 767 F.2d 1277, 1285 (9th Cir. 1984). And Parada “could hardly ask the authorities in El Salvador to certify that [he] would be persecuted [or tortured] should [he] return.' Zavala-Bonilla v. I.N.S. , 730 F.2d 562, 565 (9th Cir. 1984). Accordingly, we have rejected interpretations of immigration statutes that would mechanically require submission of a specific type of evidence in support of a noncitizen's application for relief. See id. at 567 ('[I]n considering the record as a whole, the BIA should bear in mind the difficulties [a noncitizen] encounters in providing proof of potential persecution” or torture.).

Consistently, we have cautioned that '[a]djudicating a CAT claim will generally involve speculation about the likelihood of future events, and it will not be possible, or even desirable, to quantify probabilities precisely .” Velasquez-Samayoa v. Garland , 49 F.4th 1149, 1154 n.1 (9th Cir. 2022) (emphasis added). A successful CAT claim cannot be made out without demonstrating that the Page 17government is complicit in one of the most egregious forms of abuse known to mankind—torture. Governments complicit in such actions have strong incentives to suppress and manipulate any available data on the scope of their torturous practices. See Zavala-Bonilla , 730 F.2d at 565–67; Amicus Brief of Professor Blair, at 10-15. A rigid quantitative evidence requirement ignores the reality that “[t]he data and statistical methodology that would enable a percentage to be attached to a risk of torture simply do not exist.” See Rodriguez-Molinero , 808 F.3d at 1135.

The unavailability of statistical evidence, a problem that is exasperated in countries in which torture is more-not less—likely to occur, counsels against requiring noncitizens to produce such evidence. See Velasquez-Banegas , 846 F.3d at 262 (Requiring quantitative evidence of a greater than fifty percent risk of persecution to establish eligibility for withholding of removal presents “an absurd example of line drawing, [and] it assumes unrealistically that such statistics can be computed.” Therefore, “it should be enough to entitle the applicant to [protection under CAT] if there is a substantial, albeit unquantifiable , probability that if deported he will be [tortured].” (emphasis added)). Such a requirement would eviscerate protections for the individuals who are most likely to be tortured. See Rodriguez-Molinero , 808 F.3d at 1135; Amicus Brief of Immigration Judges, at 7 (“No applicant can meaningfully quantify their personal risk of torture at the hands of clandestine security forces or within lawless detention systems where the state itself conceals the scale of its abuses. . . . CAT is meant to offer a last-resort, unconditional safeguard against torture. That purpose is impossible to reconcile with a regime in which protection turns on a judge’s ability to assign numerical probabilities to inherently unquantifiable risks.”); Amicus Brief of Refugees Page 18International, at 10 (explaining that the BIA's 'creat[ion] [of] a standard that the most at risk individuals would have particular difficulty meeting would be to nearly strike from existence the United States’s legal ratification of CAT and duly promulgated implementation regulations”).

This record well captures the concern that statistical data may be unavailable or may underestimate the scope of torturous practices in the most repressive regimes. The record reflects that the Salvadoran government “intentionally manipulat[es] data” regarding homicide rates, see Expert Report of Professor Patrick McNamara (Feb. 24, 2025) ('2025 McNamara Report'), at ¶ 56, 'releas[es] misinformation,” id. , and engages in “affirmative efforts to obscure and downplay the widespread instances of torture committed in its detention facilities,” see Amicus Brief of Refugees International, at 7. The Salvadoran government also uses “mass graves to bury the dead in state custody” to further its 'policy of withholding information about the deaths of prisoners in penal institutions.' See Cristosal, Silence is Not an Option: Research on the Practices of Torture, Death, and Failed Justice in the State of Exception (July 2024), at 11; see also Expert Report of Dr. Tommie Sue Mongomery (Feb. 21, 2025) (“Montgomery Report”), at ¶ 81 (Statistics on deaths in custody “are an extremely low count of the actual deaths inside the prisons under the [State of Exception]' because human rights organizations only report on deaths verified with the “family members of the deceased” and “independently verif[ied] by reviewing records of the national coroner's office and/or funeral homes.” The statistics therefore do not reflect bodies buried in “clandestine cemeteries and mass graves.”); Amicus Brief of Refugees International, at 6-7 (describing extrajudicial executions and “the use of mass graves to conceal deaths in Page 19detention”). Accordingly, “[i]t is practically impossible to have complete and accurate official data on arrests, homicides, femicides, prison population[s], Prosecutor's Office investigations, judicial processes, or any other activity linked to the state of exception' in El Salvador. Cristosal, supra, at 14; see also Amicus Brief of Refugees International, at 3 (“The Salvadoran government has engaged in a multi-faceted strategy to both deny the public access to information about its security forces and detention centers as well as actively hide evidence of its abuses.”).⁸

Moreover, unlike other forms of evidence, even when quantitative evidence captures broad trends of torture in the country of removal, it often will not be tailored to a noncitizen’s individual likelihood of being tortured. See 8 C.F.R. § 1208.16(c)(2) ('The burden of proof is on the applicant for withholding of removal under this paragraph to establish that it is more likely than not that he or she would be tortured if removed to the proposed country of removal.” (emphasis added)). For example, in its November 2024 decision denying Parada’s application for protection under CAT on the merits, the BIA concluded that Parada was unlikely to be tortured in a Salvadoran prison because “only a small percentage of individuals detained . . . reported harm.” In support of this conclusion, the BIA cited statistics showing that there were approximately 1,000 incidents of prisoners harmed compared with 70,000 individuals arrested during the State of Exception. Setting aside the unreliability of such statistics, those statistics shed little light on Parada’s Page 20individual likelihood of being tortured. Parada, for example, faces a higher risk of torture than others because of his unique characteristics, such as his extensive tattoos, which signify his gang affiliation, his incarceration in the United States, and his killing of a member of a rival gang—a gang that could retaliate against Parada in a Salvadoran prison.⁹ As amici put it: “The correct question is not ‘How often are Salvadoran arrestees tortured?’ but ‘What is the likelihood that someone with Mr. Parada Calderón’s background will be tortured if returned?’ That question cannot be answered by dividing two numbers. It requires reasoned judgment and consideration of the entire record.' Amicus Brief of Immigration Judges, at 13.

It is clear, therefore, that it is neither possible nor desirable to require noncitizens to produce quantitative evidence of their statistical likelihood of being tortured in the country of removal. The BIA erred by imposing a quantitative evidence requirement in this case.

Among the voluminous evidence submitted by Parada in support of his motion to reopen were three expert reports. See Montgomery Report; 2025 McNamara Report; Expert Report of Martin Flores (Feb. 24, 2025) (“Flores Report”). The BIA failed to discuss two of the expert reports, prepared by Dr. McNamara and Mr. Flores, entirely. As to the only expert report it discussed, the BIA acknowledged that “Dr. Montgomery’s report . . . described deplorable conditions of men in prison' but dismissed the specific evidence and analysis in her report on the ground that this evidence was “anecdot[al]” and “can’t substitute for quantitative evidence.” The BIA thus “failed to give reasoned consideration to the potentially dispositive testimony [of Parada's three] experts,' which requires 'remand for the agency to reconsider [his] CAT claim in light of the expert testimony.” Cole , 659 F.3d at 773.

As a threshold matter, the BIA rejected “the majority of the evidence submitted” on the ground that the evidence was “not based on new or previously unavailable evidence but reflects the continuation of abuse of suspected gang members under the state of exception since its implementation on March 27, 2022, which was prior to the respondent’s individual hearing.” See 8 C.F.R. § 1003.2(c)(1) (“A motion to reopen proceedings shall not be granted unless it appears to the Board that evidence sought to be offered is material and was not available and could not have been discovered or presented at the former hearing .' (emphasis added)). The BIA failed to identify exactly what evidence it was rejecting on this basis. Regardless, a significant amount of the evidence that Parada presented, including all three expert reports, postdated the closing of his proceeding’s administrative record in Page 22February 2024, and thus was obviously unavailable at the time of his merits hearing. As we explain in Section III.C., Parada produced sufficient material and unavailable evidence to prevail on his motion to reopen.

The Government defends the BIA’s failure to consider the expert reports. As to Mr. Flores’s report, the Government argues that Mr. Flores’s report was irrelevant because that report “focused on [Parada’s] risk of harm by Salvadoran gangs” and “the Board did not need to address [Parada’s] risk of harm from gang members in the absence of the requisite state action.'¹⁰ As to Dr. McNamara's report, the Government argues that “while the Board did not specifically cite to McNamara’s declaration, it both concluded that Parada had 'demonstrated a reasonable likelihood that he could be detained in El Salvador under the current state of exception,’ and noted that ‘the conditions of confinement in Salvadoran prisons may be difficult and below the standards we expect in the United States,’ both aspects of [Parada's] CAT claim about which McNamara opined.” The Government does not grapple with Parada’s argument that the BIA misread Dr. Montgomery’s report and made no more than a “passing mention” of her report while Page 23rejecting the core of that report on the ground that it was not quantitative evidence.

The Government’s arguments bypass the BIA’s “obligat[ion] to state reasons in the record why the testimony was insufficient to establish the probability of torture necessary to grant CAT relief.' Cole , 659 F.3d at 772 (citation modified). Its failure to “give reasoned consideration to the potentially dispositive testimony” of the experts was reversible error. Id. at 773. Here, on the basis of evolving country conditions, Dr. McNamara concluded that Parada “ would be severely injured or killed if he were forced to return to El Salvador.” 2025 McNamara Report at ¶ 134 (emphasis added). He explained that:

Under the current state of [exception], the risk to Mr. Parada's life is exacerbated by massively over-crowded prisons and indefinite periods of pretrial detention in extremely harsh prison conditions. As I state above, based on his appearance and criminal history, and my knowledge and research of current conditions, Mr. Parada will almost certainly be arrested and detained under the current state of exception. He will be beaten or worse by guards, and experience conditions that our own Department of State has confirmed are “life-threatening” and that the evidence shows are intentional. Not only that, but he will also be unprotected from attacks by Salvadoran gang members inside Page 24the prisons, where they may still exact violence with impunity.

Id. at ¶ 133. Mr. Flores's expert report concluded that Parada “would face a serious risk of harm” if removed and that he would be a 'high target to [the Mara Salvatrucha ('MS-13') and Mara 18 gangs] in El Salvador.' Flores Report at ¶¶ 26, 32. Dr. Montgomery's expert report documented widespread evidence of torture in Salvadoran prisons. She concluded that 'there is a near-100-percent certainty . . . that [Parada] will be incarcerated upon arrival” and that his 'incarceration in turn means a 100 percent certainty that he will be subjected to the mistreatment, torture, overcrowded conditions, and deprivation of food and medications” that occurs in Salvadoran prisons. Montgomery Report at ¶ 150-51.

These expert reports, which are supported by hundreds of pages of country conditions evidence, demanded more than a passing reference. The BIA cannot dismiss highly probative expert reports by stating, in a conclusory fashion, that Parada failed to establish a reasonable likelihood of meeting his burden under CAT without addressing this evidence. See Cole , 659 F.3d at 771-72; see also Castillo v. Barr , 980 F.3d 1278, 1283 (9th Cir. 2020) ('If the Board rejects expert testimony, it must state in the record why the testimony was insufficient to establish the probability of torture.” (citation modified)). The BIA erred by failing to give reasoned consideration to all three of Parada’s experts.

The record compels the conclusion that Parada established a reasonable likelihood of showing, in a reopened proceeding, that he is more likely than not to be Page 25tortured if he were removed to El Salvador. See FonsecaFonseca , 76 F.4th at 1179. Even on its plainly inadequate review of the record, the BIA concluded that Parada 'demonstrated a reasonable likelihood that he could be detained in El Salvador under the current state of exception.” It erred, however, when it considered the implications of such detention and concluded that “the evidence does not show a reasonable likelihood that [Parada] would be able to establish that it is more likely than not that he would be tortured if detained.” We conclude that the record compels the opposite conclusion because Parada has shown a “reasonable likelihood” that he would be able to show, in a reopened proceeding, that he would meet his burden under CAT. Id. The BIA thus abused its discretion by denying Parada's motion to reopen because there is a reasonable probability that, following a full merits hearing and in light of the reopened record, Parada will establish that it is more likely than not that he will be tortured if removed to El Salvador.

The administrative record in Parada’s first proceeding on the merits closed in February 2024. Parada submitted his motion to reopen one year later based on material and previously unavailable evidence that further informed his risk of being tortured if removed to El Salvador. See 8 C.F.R. § 1003.23(b)(3) (Motions to reopen must offer new evidence that is “material and was not available and could not have been discovered or presented at the former hearing.”). Importantly, the material and unavailable evidence standard is a distinct pathway, with distinct requirements, including a 90-day statutory deadline, to obtain a reopening of proceedings. Noncitizens may separately seek reopening outside of 'time and numerical Page 26limitations' based on 'changed country conditions . . . if such evidence is material and was not available and could not have been discovered or presented at the previous proceedings.' Id. § 1003.23(b)(4)(i). The availability of multiple distinct pathways to obtain reopening underscores that noncitizens seeking reopening under the first standard— the one under which Parada sought reopening-need not establish changed country conditions to prevail. See Yan Rong Zhao v. Holder , 728 F.3d 1144, 1147 (9th Cir. 2013) (The petitioner “is not required to meet the changed country conditions standard' where a motion to reopen is not numerically barred. Rather, the petitioner “need only prove that the evidence sought to be offered is material and was not available and could not have been discovered or presented at the former hearing.' (citation modified)). Accordingly, Parada must submit evidence that is (i) material, and (ii) was unavailable and could not have been discovered or presented at the time of his original proceedings. Id. ; 8 C.F.R. § 1003.23(b)(3).

The application of this standard to this record is not difficult. Parada submitted three expert declarations and voluminous country conditions evidence postdating the close of the administrative record in his original proceedings. This evidence was obviously 'unavailable' and “could not have been discovered or presented at [his] former hearing,” id. , given that the evidence did not exist at that time. The BIA's dismissal of some of Parada's new evidence on the ground that it “reflect[ed] the continuation of abuse of suspected gang members under the state of exception since its implementation on March 27, 2022, which was prior to the respondent’s individual hearing,” is incorrect as a matter of law because the BIA improperly applied the changed country conditions standard for Page 27evaluating new evidence, which does not apply to Parada’s motion to reopen. See 8 C.F.R. § 1003.23(b)(3)–(4). The only question, therefore, is whether the unavailable evidence is material. Id. We find that the evidence in this record easily clears that threshold.

Parada produced detailed evidence of his individualized risk of torture if he were removed to El Salvador under the current State of Exception.¹¹ This evidence is tailored to his unique characteristics, including that he is a former member of the Drifters gang—a rival of the two largest gangs in El Salvador: MS-13 and the 18th Street gang; was convicted of shooting at rival MS-13 gang members, for which he served Page 28nearly thirty-five years in prison in the United States, 12 during which time he became affiliated with the Mexican Mafia; and he is literally “covered” in tattoos, many of which are readily identifiable as gang tattoos.¹³

As noted, all three expert reports postdated the close of the administrative record in Parada's original hearing, in February 2024. Those reports relied on material, unavailable evidence that could not have been presented in his original hearing. For example, Dr. McNamara's analysis cited a human rights organization’s December 2024 report cataloging physical and psychological torture, including beatings that were so severe that “some [prisoners] convulsed from the beatings they received and others died.” 2025 McNamara Report at ¶ 93. That same report described prisoners 'doused . . . [by] hoses under pressure with ice- Page 29cold water,” suffocated by “tear gas,” and abused by guards using tactics such as “electric shocks, beatings . . ., confinement in the punishment cell, where there were . . . cockroaches, scorpions and mice, which caused numerous clinical cases of people contracting leptospirosis and other diseases, . . . [as well as] deprivation of . . . food, use of the bathroom[,] and . . . sunlight.” Id. at ¶ 94 (citation modified).

Other reports cited by Dr. McNamara reinforce that Salvadoran prisons are squalid, severely overcrowded, suffer from 'deplorable hygiene conditions,' and lack sufficient ventilation and water. See id. at ¶¶ 114-115; Cristosal, supra , at 24–25. As a result, disease runs rampant in the prisons. 2025 McNamara Report at ¶¶ 102–06. For example, according to a November 2024 report, the rate of tuberculosis infections in Salvadoran prisons is 72 times higher than in the general population. Id. at ¶ 102.

Based on this material evidence that was unavailable (because it did not yet exist) at Parada’s original hearing, Dr. McNamara concluded that Parada “will almost certainly be arrested and detained under the current state of exception” and that, once detained, Parada “will be beaten or worse by guards, and experience conditions that . . . are lifethreatening and that the evidence shows are intentional.' Id. at ¶ 133 (citation modified). He further found that Parada “will also be unprotected from attacks by Salvadoran gang members inside the prisons, where they may still exact violence with impunity.” Id. Accordingly, Dr. McNamara concluded that Parada “would be severely injured or killed if he were forced to return to El Salvador.” Id. at ¶ 134.

Dr. Montgomery similarly provided new, material, and previously unavailable evidence of the torturous conditions Page 30that Parada would face if removed to El Salvador. Her report relied on research she conducted in El Salvador through November 2024. See Montgomery Report at ¶ 14. For example, her report contains first-hand accounts of torture in Salvadoran prisons, including of beatings, starvation, and other forms of torture. One prisoner, who Dr. Montgomery interviewed in November 2024, was so deprived of food in prison that he lost one hundred pounds—more than fortyfive percent of his body weight—in just six months. Id. at ¶ 71. That same prisoner described being subjected to and witnessing horrific acts of torture and murder, including seeing twenty-five men die in twenty-two days in a prison in Izalco. Id. at ¶¶ 75, 81. Another detainee, interviewed in August 2024, described witnessing thirty-four or thirty-five men die over a five-month period in a Salvadoran prison, noting that “[n]o one died a natural death.” Id. at ¶ 81. A third prisoner, who provided an account of his experience in prison in June 2024, witnessed deaths “every day” in prison. Id.

Based on material evidence that was unavailable (because it did not yet exist) at the time of Parada’s original hearing, Dr. Montgomery concluded that “there is a near100-percent certainty . . . that [Parada] will be incarcerated on arrival” and that his “incarceration in turn means a 100 percent certainty that he will be subjected to the mistreatment, torture, overcrowded conditions, and deprivation of food and medications” that occurs in Salvadoran prisons. Id. at ¶ 150-51. Dr. Montgomery concluded that the state imposes torturous conditions, “carried out by prison officials,” as a matter of “state policy . . . for the purpose of tormenting the prisoners in . . . custody.' Id. at ¶ 152. In particular, Parada would be 'near[ly] certain[] . . . [to] experience violence such as Page 31beatings by prison guards and gang members, waterboarding, electrocution, stress positions, and other acts of torture,' as well as 'other harsh and life-threatening conditions including overcrowding, unhygienic conditions, and lack of food, all of which are intended by the government to prolong the suffering endured by those incarcerated under the State of Emergency.” Id. at ¶ 159. Even if Parada was lucky enough to be the “rare” prisoner who is released, Dr. Montgomery concluded that his “tattoos” and criminal history “would virtually guarantee his re-arrest.” Id. at ¶ 153. Lastly, Dr. Montgomery concluded that Parada faced “distinct risks of harm . . . by MS-13 and 18th Street [gang members] either inside the prisons or outside of the prisons.' Id.

Finally, Martin Flores, an expert on criminal gangs, provided a report focused on Parada’s risk of harm at the hands of Salvadoran gang members. Flores’ expert report was prepared based on evidence that was current through the date of the report, February 24, 2025, as well as an interview with Parada conducted that same month. Flores concluded that Parada would be “considered a high value target” by the “MS-13 or 18th Street gangs,” and that he would “face a serious risk of harm” as a result. Flores Report at ¶ 6, 32. Flores explained that Parada’s association with the Drifters and Mexican Mafia would be 'obvious' because of his accent and tattoos. Id. at ¶ 6. He explained that Parada's conviction for murder of an MS-13 gang member and association with rival gangs also “make him a high target” to these gangs. See id. at ¶¶ 26-32. Despite the fact that MS-13 was not “able to exact revenge while [Parada] was in prison due to the politics in California prisons,” MS-13 gang members “would be able and motivated to do so in settings not controlled by [the Mexican Mafia], including in El Page 32Salvador.” Id. at ¶ 30 (“MS-13 cliques in El Salvador carry out . . . hits on individuals for acts carried out against cliques in the U.S.”).

Parada also submitted material, previously unavailable country conditions evidence, including a July 2024 report prepared by Cristosal, a human rights organization, reflecting dismal conditions in El Salvador. The Cristosal report determined, based on a study of 1,178 prisoners detained between 2022 and 2024, that in 'no case was evidence presented to prove that the prisoners were committing or had just committed the crime,” that “arrest reports or requests for provisional detention submitted by the Prosecutor General’s Office did not contain direct evidence, testimony, or sufficient evidence to justify the charges,” and that “in none of the cases did the Prosecutor’s Office present evidence of the prisoners’ links to a gang.” See Cristosal, supra , at 10. That same report determined that 'there is ample evidence that the state apparatus responsible for implementing the state of exception has caused serious and massive human rights violations.” Id. at 12. Cristosal, as well as other human rights organizations, also documented detainee deaths. Id. at 9. As of May 2023, just over one year into the State of Exception, Cristosal had documented 160 detainee deaths in prison. Montgomery Report at ¶ 80. By January 2025, another human rights organization had documented 366 deaths in custody (a rate of more than one prisoner death every three days). Id. These statistics represent “an extremely low count of the actual deaths inside the prisons' given the aforementioned difficulties with obtaining reliable statistical evidence of extrajudicial torture and other abuses. See id. at ¶ 81.

The record reflects that Salvadoran officials directly perpetrate and acquiesce to widespread acts of torture and Page 33other abuses in Salvadoran prisons, including beatings, starvation, electric shocking, sexual assaults, and the denial of medical treatment to prisoners. See, e.g., id. at ¶¶ 71-79. As the expert reports and country conditions evidence make clear, human rights abuses, up to and including torture and extrajudicial killings, are the modus operandi of Salvadoran prisons. To be sure, CAT requires an individualized inquiry, and we do not hold that all deportees to El Salvador will meet their burden of establishing a likelihood of future torture upon removal. Parada's individualized evidence of his likelihood of future torture, including expert reports considering his unique characteristics—such as his criminal history, association with multiple international gangs that are rivals of Salvadoran gangs, and tattoos—is sufficient to establish prima facie eligibility for protection under CAT. The BIA thus abused its discretion by denying his motion to reopen.

* * *

Our dissenting colleague makes several compounding errors in concluding that the BIA did not impose a quantitative evidence requirement on Parada, did not fail to consider his expert evidence, and that Parada did not meet his burden of establishing a reasonable likelihood that he would prevail in a reopened proceeding. To begin, the dissent concedes that “[i]t is hard to argue with the general statement that quantitative evidence is helpful but not required,” but insists that the BIA, in citing Benedicto , was doing no more than “commenting on the evidentiary weight of Parada’s evidence and explaining that his case may have been aided if he had contributed quantitative evidence.' See Dissent at 46. Curiously, it is the dissent's dissection of Page 34Benedicto that unmasks the quantitative evidence requirement that the dissent claims does not exist in this case. In the dissent's telling, Benedicto recognized that “anecdotes from current and former detainees . . . are hard to quantify and therefore are typically not sufficient to establish” a petitioner’s likelihood of torture. Id. (emphasis added). Attempting to thread the needle, the dissent suggests that the BIA, in applying Benedicto , meant to say only that Parada's case would be 'aided if he had contributed quantitative evidence.' Id. But that is not what the BIA said. Nor is it what Benedicto said-there, we denied Benedicto's petition because the evidence did not “establish that Benedicto himself would ‘more likely than not’ be tortured on removal.” Benedicto v. Garland , 12 F.4th 1049, 1065 (9th Cir. 2021) (emphasis added).

The dissent cannot have it both ways. The core function of an immigration judge is to weigh the evidence before him and determine whether, based on a holistic review of the record, it is more likely than not that a noncitizen will be tortured with the consent or acquiescence of the government if he is removed to his country of origin. Anecdotal evidence cannot be rejected simply because it would be more expedient or less burdensome for the immigration judge to decide the case based on statistical evidence. The dissent offers no response to our textual analysis of CAT's implementing regulations. See 8 C.F.R. § 1208.16(c)(3) (“[A]ll evidence relevant to the possibility of future torture shall be considered.”). Instead, it insists that the BIA has not done exactly what the agency said that it did: reject strong evidence of Parada’s likelihood of future torture because the evidence took a particular form-documentary evidence, rather than statistical evidence.

Page 35The dissent’s misreading of the BIA’s decision leads to its most fundamental error: the dissent concludes that Parada has not met his burden because '[t]he record does not compel [the conclusion] . . . that anywhere close to half of the detainees in El Salvador are tortured with government consent or acquiescence, which is what Parada had to prove to be entitled to CAT protection.” 14 See Dissent at 43. The dissent misrepresents the core function of CAT: a petitioner's ability to obtain protection depends on his individual likelihood of being tortured if he were to be removed. See 8 C.F.R. § 1208.16(c)(2). As we have explained, the question is not whether half of all detainees would be tortured if removed to El Salvador. That analysis is irrelevant to Parada’s risk of being tortured. The proper question is whether a noncitizen with Parada's individual characteristics and circumstances is reasonably likely to be able to show, in a reopened proceeding, that he is more likely than not to be tortured if removed to El Salvador. See Ridore , 696 F.3d at 912 ('[T]he key question[] . . . [is] whether [the noncitizen] has shown individual circumstances in his case that there is a likelihood that he will face torture qualifying him for CAT.' (internal quotation marks and citation omitted)).

Like the BIA, the dissent misses this distinction and then doubles down by quoting misleading statistics concerning the number of in-custody deaths in El Salvador.¹⁵ By Page 36dismissing Parada’s voluminous and individualized evidence of his risk of torture as merely anecdotal, the dissent follows the well-trodden path of error blazed by many immigration judges in recent years. See Amicus Brief of Immigration Judges, at 1. Today's decision rights the course.

The dissent's suggestion that the BIA considered all three of Parada's expert reports is wrong. The dissent concedes, as it must, that the BIA failed to address the expert reports of two of Parada's three experts entirely. See Dissent at 48–51. The dissent makes various excuses for the BIA, including that it is overworked, protected by the presumption of regularity, and cited to Parada’s motion to reopen which itself had cited to Parada's expert reports. Id. We acknowledge the large backlog of immigration cases and the heavy workload that judges of the BIA face. See Dissent at 51. But that is no excuse for ignoring 'highly probative [and] potentially dispositive evidence.” See Cole , 659 F.3d at 772. Our binding precedent demands that the BIA explain why it is rejecting such evidence. Id. Contrary to the suggestion of the dissent, the presumption of regularity cannot overcome this binding precedent, see id. , or the statutory requirement that the BIA consider 'all evidence relevant to the possibility of future torture.'⁸ C.F.R. § 1208.16(c)(3). The dissent is thus left with only the unavailing arguments that the BIA cited to Parada’s motion to reopen in denying the motion, and that Parada’s motion itself had cited to his expert reports. See Dissent at 51. But hard as we squint, we cannot find any “reasoned Page 37consideration' in the BIA's passing citations to Parada's motion to reopen in light of its complete failure to address the specific evidence in the expert reports. See Cole , 659 F.3d at 772. Our precedent simply demands more. Id.

The dissent errs, as did the BIA, in suggesting that much of the evidence that Parada offered in support of his motion to reopen is irrelevant because the evidence “is duplicative of the evidence he had already presented to the IJ” and is 'not material considering the prior record in [Parada's] earlier immigration proceedings.' See Dissent at 53. As we have already explained, two distinct standards apply to motions to reopen. Under the first standard, a motion to reopen is granted where the noncitizen produces evidence that is “material and was not available and could not have been discovered or presented at the former hearing.” See 8 C.F.R. § 1003.23(b)(3). The second standard applies to motions to reopen brought after the statutory deadline for filing a motion or where a motion to reopen has already been brought. See 8 C.F.R. § 1003.23(b)(4). Under the latter standard, the noncitizen must show 'changed country conditions arising in the country of nationality or the country to which removal has been ordered.” Id. at § 1003.23(b)(4)(i).

Here, the motion to reopen was brought under 8 C.F.R. § 1003.23(b)(3). Unlike motions to reopen brought under subsection (b)(4), the BIA does not undertake a comparative analysis of the evidence presented in support of the motion to reopen and the evidence presented in the original proceeding when considering a motion to reopen filed under subsection (b)(3). Because Parada submitted material evidence that was unavailable at the time of his original Page 38removal proceedings, the BIA erred in rejecting the evidence.¹⁶

Lastly, we reject any suggestion that a 'belief that El Salvador’s State of Exception is ‘an extreme crackdown on civil liberties' . . . color[ed] [our] view of the evidence.' See Dissent at 53 n.3. The dissent concedes, as it must, that detainees are subject to “admittedly harsh treatment.” See Dissent at 53. In so doing, the dissent neglects to mention some of the most vile acts of torture-including sexual assaults, starvation, and extrajudicial executions—that occur in Salvadoran prisons. See id. (referencing only 'beatings, electric shocks, and tear gas being used on detainees”). But the dissent does not deny that prisoners endure such torture in these prisons. The record overflows with evidence of torture perpetrated by Salvadoran officials against prisoners. We conclude, with little difficulty, that Parada met his burden of establishing his “reasonable likelihood” of succeeding on the merits of his application for relief in a reopened proceeding. Fonseca-Fonseca , 76 F.4th at 1181. That decision rests on the record evidence, not the dissent’s speculation.¹⁷

Page 39Parada submitted extensive evidence of the overwhelming likelihood that he, based on his individual characteristics, would face torture if he were removed to El Salvador. The BIA abused its discretion by denying his motion to reopen in multiple respects. First, it imposed an unprincipled and atextual “quantitative evidence” requirement that stands in direct conflict with the plain text of CAT’s implementing regulations. See 8 C.F.R. § 1208.16(c)(3). Second, it erred by failing to consider “highly probative or potentially dispositive evidence” in the record, including expert declarations. See Cole , 659 F.3d at 772. Because the record compels the conclusion that Parada met his burden of establishing prima facie eligibility for relief, see Fonseca-Fonseca , 76 F.4th at 1181, we GRANT his petition for review and REMAND to the BIA with instructions to: (1) grant Parada's motion to reopen his removal proceedings, and (2) conduct a full hearing on the merits of Parada's application for protection under CAT.

PETITION GRANTED. 18Page 40 BEA, Circuit Judge, dissenting:

The majority purports to publish today's opinion to correct “the increasingly widespread misapplication of plain and unambiguous legal standards by our immigration courts.” Maj. Op. at 6. Unfortunately, it is the majority that misapplies plain and unambiguous legal standards.

We “review[] the denial of a motion to reopen for abuse of discretion, with broad deference to the agency’s decision.' Tzompantzi-Salazar v. Garland , 32 F.4th 696, 702 (9th Cir. 2022). Yet, despite our highly deferential standard of review, the majority concludes that the BIA abused its discretion because it 'imposed an unprincipled and atextual ‘quantitative evidence’ requirement.” Maj. Op. at 39. In fact, the BIA did nothing more than recite verbatim a passage from our decision in Benedicto v. Garland , 12 F.4th 1049 (9th Cir. 2021), which discusses the relative evidentiary weight of quantitative and anecdotal evidence. For the majority, the BIA’s accurate recitation of uncontroversial legal principles can amount to reversible error.

That conclusion is somewhat absurd. Read in its proper context, it is clear that the BIA did not ignore Petitioner Alfredo Parada Calderon’s (“Parada”) anecdotal evidence of torture by requiring or considering only statistical evidence-the majority's alleged 'quantitative evidence requirement.” Maj. Op. at 7. Instead, the BIA considered all the evidence and permissibly concluded that Parada's evidence did not establish a reasonable likelihood that he would be able to establish prima facie eligibility for deferral of removal under the Convention Against Torture (“CAT”).

Page 41As a three-judge panel, we have no authority to overrule Benedicto . So, it cannot be the case that the BIA abuses its discretion when it correctly adheres to that precedent. Therefore, I respectfully dissent.

I begin by providing some information about Parada that the majority opinion neglects to mention. Parada, a native and citizen of El Salvador, entered the United States unlawfully in 1980 and was accorded lawful permanent resident status in February 1990. When Parada was fifteen years old, he was 'heavily involved' with a Los Angeles street gang called the “Drifters 23 Malos.”

In August 1990, when he was sixteen years old, he shot and killed a twelve-year-old girl who was outside an El Pollo Loco Restaurant. She was on her way from school with one of her friends to meet up with some teenage boys at the El Pollo Loco. Unfortunately for her, the El Pollo Loco was in Drifters territory. Parada and a fellow Drifter approached the teenagers to “ask them where they were from.” Parada and his gang associate “screamed, ‘Drifters, our neighborhood.’” When some of the teenagers yelled back, “Mara Salvatrucha MS,'¹ Parada “pulled out [his] handgun . . . and open fired.” The twelve-year-old girl was fatally 'shot in the back of her head.” The friend with whom she had walked from school hid “behind the car covering herself,” while Parada and his associate “chased the [boys] across the alleyway.” Two of the fleeing teenagers were also shot. After the unprompted shooting, Parada and his associate went to the arcade to play videogames. Their play Page 42was interrupted when police officers showed up to question them.

Parada was convicted of one count of murder and three counts of attempted murder.² He was sentenced to 34 years and eight months to life in prison with the possibility of parole. While in prison, Parada joined and became a leader of the Mexican Mafia, but later disassociated from the gang. Parada was paroled from prison in 2023 and was immediately placed in U.S. Immigrations & Customs Enforcement custody, where he remains today.

The Department of Homeland Security commenced removal proceedings against Parada in October 2023. Although Parada contested the charges of removability, the Immigration Judge ('IJ') sustained them. Thereafter, Parada applied for asylum, withholding of removal, and protection under the CAT. The IJ denied all forms of relief. Parada appealed the IJ’s denial of his application for CAT protection. The BIA reversed the IJ and remanded the record to the IJ for the IJ to make additional findings of fact based on the evidence already in the record that the IJ had failed properly to consider. On remand, the IJ again denied Parada all forms of relief. The BIA affirmed because it found no clear error in the IJ’s factual or legal conclusions. Parada filed a petition for review of the BIA’s decision in this Court.

While his petition was pending before this Court, Parada filed a motion to reopen with the BIA, seeking to submit additional evidence of torture in El Salvador. The BIAPage 43 denied Parada's motion to reopen because Parada's 'new evidence, when considered together with the evidence presented at the hearing, does not establish prima facie eligibility for deferral of removal under the CAT.” The BIA, quoting Benedicto , 12 F.4th at 1064–65, noted that, “[w]hile we do not condone the treatment described in the evidence presented, '[a]necdotes-no matter how disturbing-can't substitute for quantitative evidence, which is what CAT requires.’” Parada filed a timely petition for review.

We review the BIA’s denial of a motion to reopen for abuse of discretion. Najmabadi v. Holder , 597 F.3d 983, 986 (9th Cir. 2010). We 'defer to the BIA's exercise of discretion unless it acted arbitrarily, irrationally, or contrary to law.” Id. “We review the BIA’s determination of purely legal questions de novo, and review its factual findings for substantial evidence.' Id. The substantial evidence standard requires that we uphold the BIA’s factual findings if they are “supported by reasonable, substantial, and probative evidence on the record.” Cordon-Garcia v. I.N.S. , 204 F.3d 985, 990 (9th Cir. 2000) (simplified).

Under our highly deferential standard of review, Parada’s petition should be denied because the BIA considered all evidence relevant to the possibility of torture and reasonably concluded that Parada “has not shown prima facie eligibility for protection under the CAT.” The record does not compel a contrary conclusion because there is no evidence that anywhere close to half of the detainees in El Salvador are tortured with government consent or acquiescence, which is what Parada had to prove to be entitled to CAT protection.⁸ C.F.R. § 1208.16(c)(2); see Tzompantzi-Salazar , 32 F.4th at 706-07 (holding thatPage 44 “generalized evidence would need to show” that the risk of torture is great enough such that over half of the subject population is at risk).

The majority disagrees because it asserts that the BIA “imposed an unprincipled and atextual ‘quantitative evidence’ requirement that stands in direct conflict with the plain text of CAT’s implementing regulations.” Maj. Op. at 39. The majority concludes that, “[u]nder the correct legal standards, the record evidence compels the conclusion that Parada met the burden of showing that he has a ‘reasonable likelihood of [succeeding] on the merits’ of his application in a reopened proceeding.' Maj. Op. at 7 (quoting FonsecaFonseca v. Garland , 76 F.4th 1176, 1181 (9th Cir. 2023)). But, as I explain below, the BIA imposed nothing close to the 'quantitative evidence requirement' that the majority imagines.

The majority asserts that the BIA committed numerous errors due to it wrongly imposing a quantitative evidence requirement when it quoted our decision in Benedicto . Maj. Op. at 8-10. To untangle the many errors the majority alleges, one must first understand our decision in Benedicto .

In Benedicto , we reasoned that “[a]necdotes—no matter how disturbing—can’t substitute for quantitative evidence, which is what CAT requires.'¹² F.4th at 1065. We affirmed the BIA's dismissal of a Dominican petitioner's appeal of his application for CAT protection because the record evidence did not compel the inference that the Dominican government would intentionally torture him were he to be arrested after he was removed from the United States. Id. at 1053, 1063. As to the petitioner's risk of torture in Dominican Republic prisons, we noted that Page 45although the evidence showed that “some torture occurs” in Dominican Republic prisons, “nothing in the record quantifie[d] the risk of that torture or demonstrate[d] any regularity of such activity.' Id. at 1065 (simplified). Further, we noted that the Dominican government had instituted a program to improve the health of prisoners and permitted visits by nongovernment observers. Id. Therefore, we held that the petitioner’s “[s]imply pointing to evidence of instances of torture, without more,' was not sufficient to “establish that [he] himself would ‘more likely than not’ be tortured on removal.” Id.

Returning to the present case, it is clear that Benedicto is applicable. Parada’s motion to reopen presented additional evidence of the harsh prison conditions in El Salvador. However, the BIA noted that Parada’s evidence consisted of anecdotes from current and former detainees, which, as we recognized in Benedicto , are hard to quantify and therefore are typically not sufficient to establish that a petitioner would more likely than not be tortured on removal. Benedicto , 12 F.4th at 1065. Thus, the BIA, quoting Benedicto , 12 F.4th at 1064–65, stated that “[w]hile [it] [did] not condone the treatment described in the evidence presented, '[a]necdotes-no matter how disturbing-can't substitute for quantitative evidence, which is what CAT requires.’”

It is this innocuous line from the BIA’s decision that the majority relies on to conclude that the BIA 'imposed an unprincipled and atextual ‘quantitative evidence’ requirement.” Maj. Op. at 39. Read less polemically, it is clear the BIA-and this Court in Benedicto —did not impose on petitioners seeking CAT relief a “quantitative evidence requirement,” in the sense of requiring or considering only statistical evidence of torture, as the majority ascribes to that Page 46phrase. Rather, the BIA was simply commenting on the evidentiary weight of Parada’s evidence and explaining that his case might have been strengthened by more quantitative evidence of torture, rather than anecdotal accounts from current and former detainees. Nor was the BIA rejecting anecdotal evidence merely because doing so was “expedient.” Contra Maj. Op. at 34. Therefore, the BIA did not abuse its discretion by acting contrary to law, and we should not have granted Parada’s petition for review on that basis.

As for the merits of Parada’s motion to reopen, the BIA concluded that Parada’s “evidence does not show a reasonable likelihood that he would be able to establish that it is more likely than not that he would be tortured if detained' in El Salvador. See Fonseca-Fonseca , 76 F.4th at 1179. Therefore, the BIA denied his motion to reopen because he failed to carry his burden of establishing “prima facie eligibility for deferral of removal under the CAT.”

The majority disagrees for a variety of reasons. But each reason stems from the majority’s erroneous conclusion that the BIA committed legal error by imposing a quantitative evidence requirement and thereby ignoring anecdotal evidence. As discussed above, the “quantitative evidence requirement” supposedly imposed on Parada exists only in the majority’s imagination. Therefore, the majority disagrees with the BIA because it 'simply reached a different overall conclusion.' See Hernandez v. Garland , 52 F.4th 757, 771 (9th Cir. 2022). Such a disagreement does not mean that the BIA’s conclusion was not supported by substantial evidence, and indeed, the BIA’s conclusion here was so supported.

Page 47The majority asserts that “[t]he BIA rejected 'the majority of the evidence' that Parada submitted on the grounds that this evidence was 'not based on new or previously unavailable evidence[.]’” Maj. Op. at 9. Although the majority cannot tell which evidence the BIA rejected, it takes as a given that the BIA rejected some evidence. Maj. Op. at 21. Yet, read in context, the BIA was not refusing to consider Parada’s evidence—it considered all of it. Instead, the BIA was simply articulating that, as the majority recognizes, “[a] motion to reopen proceedings shall not be granted unless it appears to the [BIA] that evidence sought to be offered is material and was not available and could not have been discovered or presented at the former hearing.” 8 C.F.R. § 1003.2(c)(1); see Maj. Op. at 21.

The majority forgets that “[w]e have long recognized a presumption that the agency reviewed all relevant evidence submitted to it.” Cruz v. Bondi , 146 F.4th 730, 739 (9th Cir. 2025); see Ross and Morrison v. Reed , 14 U.S. 482, 486 (1816) ('It is a general principle to presume that public officers act correctly until the contrary be shown.”). Therefore, it is Parada’s burden to establish that the BIA did not consider all relevant evidence. Cruz , 146 F.4th at 739. That burden is 'heavy, demanding that a petitioner present ‘clear, affirmative evidence’ that the agency did not review the evidence before it.” Id. at 740 (quoting Gov’t of Guam v. Guerrero , 11 F.4th 1052, 1060 (9th Cir. 2021)). To satisfy this heavy burden, Parada must show what “evidence in particular was supposedly ignored” and “must explain why that evidence was not just material, but ‘highly probative or potentially dispositive.’” Id. (quoting Cole v. Holder , 659 F.3d 762, 772 (9th Cir. 2011)) (emphasis in original). Parada has not done so.

Page 48Parada argues that the BIA 'ignored evidence of widespread torture in Salvadoran prisons.' Parada cites admittedly harsh conditions in Salvadoran prisons, including beatings by prison guards. Yet, the BIA considered such evidence when it earlier affirmed the IJ’s denial of Parada’s application for CAT protection. Thus, Parada has not presented new or previously unavailable evidence of torture; he has merely rehashed his disagreement with the BIA's earlier decision. See Iturribarria v. I.N.S. , 321 F.3d 889, 895 (9th Cir. 2003) (“a motion to reopen may only be granted upon a proffer of material evidence that was not available and could not have been discovered or presented at the former hearing” (simplified)).

As for new evidence, Parada asserts that the BIA ignored evidence of El Salvador's 'massively overcrowded cells without access to sunlight where they suffer malnutrition and disease.' However, the BIA 'acknowledge[d] that the conditions of confinement in Salvadoran prisons may be difficult and below the standards we expect in the United States.” Again, Parada simply restates his disagreement with the BIA. Therefore, Parada has not met his heavy burden to rebut the presumption of regularity. Cruz , 146 F.4th at 740.

Nowhere is the majority’s disregard of our modest role when reviewing decisions of the BIA more apparent than its conclusion that the BIA abused its discretion “by failing to give reasoned consideration to all three of Parada’s experts.” Maj. Op. at 24. The majority, in a role better suited for Parada’s attorney, criticizes the Government for defending the BIA’s decision. Maj. Op. at 22–23. The majority asserts that Parada's 'expert reports, which are supported by Page 49hundreds of pages of country conditions evidence, demanded more than a passing reference.” Maj. Op. at 24.

That the majority would have done things differently if it were the BIA is not a basis to grant Parada’s petition for review. The BIA need not “individually identify and discuss every piece of evidence in the record.' Hernandez , 52 F.4th at 770. In fact, all we ask is that the BIA “consider the issues raised, and announce its decision in terms sufficient to enable a reviewing court to perceive that it has heard and thought and not merely reacted.” Najmabadi , 597 F.3d at 990 (quoting Lopez v. Ashcroft , 366 F.3d 799, 807 n.6 (9th Cir. 2004)). The BIA did that here.

The majority asserts that 'the BIA committed the paradigmatic error that we warned against in Cole v. Holder by failing to give reasoned explanation to the three expert reports offered by Parada.” Maj. Op. at 11. The majority reasons that this case is comparable to Cole because '[t]he BIA failed to discuss two of [Parada’s] expert reports” and rejected “the only expert report it discussed . . . on the ground that this evidence was ‘anecdot[al]’ and ‘can’t substitute for quantitative evidence.'' Maj. Op. at 21. Not only does the majority persist in its misreading of the BIA's quoting Benedicto , but the majority’s comparison to Cole is inapt.

In Cole , we granted a petition for review, in part, “[b]ecause the BIA failed to give reasoned consideration to potentially dispositive testimony by [the petitioner’s] expert witnesses.' 659 F.3d at 764. The petitioner in Cole supported his CAT claim with testimony from two experts. Id. at 766. The IJ denied the petitioner's application for CAT protection. Id. at 769. The IJ 'faulted the expert testimony . . . as lacking evidence that individuals similarly situated to [petitioner] had been incarcerated and tortured.” Id. The Page 50BIA affirmed and discounted the testimony of one of the petitioner’s experts “on the ground that the expert failed to give specific examples corroborating his opinion.” Id. We remanded to the agency because we determined that the BIA had “failed to give reasoned consideration to the potentially dispositive testimony of [petitioner’s] two experts.” Id. at 773. We noted that “the BIA mischaracterized the record with regard to one of the expert’s consistency with the State Department reports, criticized that expert’s testimony on a basis belied by the record, and failed even to acknowledge [petitioner's] other expert witness.' Id. In light of such overwhelming errors by the BIA, we determined that the BIA “failed to give reasoned consideration to the potentially dispositive testimony of [petitioner's] two experts.' Id.

In contrast, here, the BIA’s decision regarding Parada’s motion to reopen evidences that the BIA considered Parada’s expert reports. The BIA discussed Dr. Montgomery’s expert report in-depth. As the majority recognizes, 'the BIA acknowledged that ‘Dr. Montgomery’s report . . . described deplorable conditions of men in prison.’” Maj. Op. at 21. The BIA referenced the statement of “a former prisoner that all prisoners are subjected to beatings upon arrival.' However, as for Parada’s risk of torture, the BIA noted that a police chief told Dr. Montgomery 'that the police no longer had a daily quota of arrests to meet.” But, weighing Parada's other evidence, the BIA concluded that Parada “ha[d] demonstrated a reasonable likelihood that he could be detained in El Salvador under the current state of exception.” As for Parada’s other experts, the BIA acknowledged that Parada “submit[ed] . . . evidence of country conditions . . . that document intentionally cruel and torturous prison conditions and the stigmatization of tattoos and prior criminal history by the government.' The BIA cited the Page 51sections of Parada’s motion to reopen, which included Dr. Patrick McNamara's and Martin Flores's expert reports. Therefore, unlike Cole , there is no indication that the BIA failed to consider Parada’s experts.

In disregarding the presumption of regularity, the majority also ignores the realities of immigration proceedings. The judges of the BIA are often overworked, tasked with deciding an ever-growing backlog of appeals. See Hillel Smith & Holly Straut-Eppsteiner, CONG. RSCH. SERV., IG10022, IMMIGRATION COURT PROCEEDINGS: PROCESS AND DATA (2025), https://www.congress.gov/crsproduct/IG10022. Recognizing these circumstances, we do not require the BIA to 'discuss each piece of evidence submitted,' so long as its decision indicates that it has considered all the evidence. Cole , 659 F.3d at 771. We have favorably quoted the Second Circuit’s decision in Wang v. BIA , which 'reject[ed] any implication . . . that where the BIA has given reasoned consideration to the petition, and made adequate findings, it must expressly parse or refute on the record each individual argument or piece of evidence offered by the petitioner.' Najmabadi , 597 F.3d at 990 (quoting Wang v. BIA , 437 F.3d 270, 275 (2d Cir. 2006)). My colleagues may demand more from the BIA, but our precedent does not. The BIA more than adequately fulfilled its obligations and did not abuse its discretion when it considered Parada’s expert reports but ultimately decided to deny his motion to reopen.

Based on its thorough review of the record, the BIA concluded that Parada's 'new evidence, when considered together with the evidence presented at the hearing, does not establish prima facie eligibility for deferral of removal under Page 52the CAT.” The record does not compel a contrary conclusion because there is minimal evidence quantifying the extent of torture to compel a finding that anywhere close to half of the detainees in El Salvador are tortured with government consent or acquiescence. See TzompantziSalazar , 32 F.4th at 706–07.

The BIA concluded that Parada “demonstrated a reasonable likelihood that he could be detained in El Salvador under the current state of exception.” But the BIA determined that Parada’s evidence did not show that it was “more likely than not that he would be tortured if detained” because the “difficult” “conditions of confinement in Salvadoran prisons” do not amount to torture. That is correct because harsh conditions of confinement, absent intent, do not amount to torture. See Villegas v. Mukasey , 523 F.3d 984, 989 (9th Cir. 2008) (holding that terrible conditions in Mexico's mental health system did not amount to torture because Mexican officials took steps to improve conditions); Andrade v. Garland , 94 F.4th 904, 915 (9th Cir. 2024) (holding that terrible conditions in Mexico’s mental health facilities did not constitute torture because “the record also support[ed] a plausible inference that budgetary constraints . . . [were] to blame'). Further, the number of in-custody deaths is small compared to the number of detainees in El Salvador, i.e., 265 in-custody deaths of around 79,000 Salvadoran prisoners. The rest of the evidence Parada submitted consists of anecdotes, which, as discussed above, are typically not sufficient to establish that a petitioner would more likely than not be tortured on removal. Benedicto , 12 F.4th at 1065. Therefore, the BIA’s conclusion that Parada’s evidence does not establish prima facie eligibility for relief is supported by reasonable, substantial, and probative evidence on the record.

Page 53The majority disagrees because '[t]he record compels the conclusion that Parada established a reasonable likelihood of showing, in a reopened proceeding, that he is more likely than not to be tortured if he were removed to El Salvador.” Maj. Op. at 24–25. With respect, it does not.

As discussed above, to qualify for reopening, Parada must offer evidence that is “material and was not available and could not have been discovered or presented at the former hearing.” Yan Rong Zhao v. Holder , 728 F.3d 1144, 1147 (9th Cir. 2013) (quoting 8 C.F.R. § 1003.2(c)(1)). The majority simply takes it as a given that Parada offered unavailable evidence because he 'submitted three expert declarations and voluminous country conditions evidence postdating the close of the administrative record in his original proceedings.” Maj. Op. at 26. Yet, upon review of the record, Parada’s offered evidence is duplicative of the evidence he had already presented to the IJ. For example, Dr. McNamara's 2025 expert report largely restates the information in his 2023 report, which was considered during Parada’s earlier immigration proceedings. Nor is Parada’s offered evidence material considering the prior record in his earlier immigration proceedings.

The majority cites admittedly harsh treatment of detainees during El Salvador’s State of Exception, 3 including instances of beatings, electric shocks, and tear gas Page 54being used on detainees. Maj. Op. at 28-29. Yet, “[t]orture is an extreme form of cruel and inhuman treatment and does not include lesser forms of cruel, inhuman or degrading treatment or punishment that do not amount to torture.' Garcia v. Wilkinson , 988 F.3d 1136, 1147–48 (9th Cir. 2021) (quoting 8 C.F.R. § 1208.18(a)(2)). Even extensive beatings by police officers do not compel a finding of torture. See Ahmed v. Keisler , 504 F.3d 1183, 1188-89, 1201-02 (9th Cir. 2007) (concluding that petitioner being beaten and jailed on three occasions by police did not justify CAT relief). The majority also selectively cites poor and unsanitary prison conditions. Maj. Op. at 29-31. The majority quotes Dr. McNamara’s assertion that such conditions are intentional, Maj. Op. at 29, but does not cite that Dr. McNamara reported that the government of El Salvador had increased spending per inmate following an initial reduction during the first year of the State of Exception. The majority also ignores evidence that “conditions for prisoners [had] improved slightly, specifically in terms of food and overcrowding,' at some Salvadoran prisons since 2023. Further, the majority ignores the 2023 Department of State Human Rights Report, which found that 'the government took credible steps to identify and punish officials who may have committed human rights abuses.' See Kazlauskas v. I.N.S. , 46 F.3d 902, 906 (9th Cir. 1995) (giving strong evidentiary weight to the Department of State country report, describing it as 'the most appropriate and perhaps the best resource” on country conditions (simplified)). As for any quantitative evidence that undercuts its view, the majority discounts this evidence because El Salvador has a 'strong incentive[] to suppress and manipulate any available data on the scope of [its] torturous practices.' Maj. Op. at 17; see id. at 19-20Page 55 (discounting low number of injuries in Salvadoran prisons because the statistics are “unreliabl[e]”). It is difficult to see how the majority's review of the evidence can be called deferential review at all.

At bottom, the record contains some evidence that Parada might be tortured, but also evidence that he will not be, given that El Salvador has taken steps to improve prison conditions. See Villegas , 523 F.3d at 989. Therefore, “we must uphold the [BIA’s] decision” because the record does not compel a contrary conclusion. Castillo v. Barr , 980 F.3d 1278, 1283 (9th Cir. 2020).

Today, my colleagues break new ground by suggesting “that Salvadoran officials directly perpetrate and acquiesce to widespread acts of torture and other abuses in Salvadoran prisons.” Maj. Op. at 32–33. Prior to today’s decision, our Court has repeatedly upheld denials of CAT applications based on similar State of Exception evidence and arguments that Parada made below and to this Court, doing so in unpublished decisions without any noted dissents. See Trejo-Campos v. Bondi , No. 24-3901, 2025 WL 1473797, at *1–2 (9th Cir. 2025); Rivera v. Bondi , No. 24-1380, 2025 WL 927187, at *2 (9th Cir. 2025); Chavez v. Garland , No. 23-251, 2024 WL 490350, at *2 (9th Cir. 2024). Yet, as Judge VanDyke has warned, today’s majority is permitted to reach its conclusion, unmoored from precedent, because these decisions that properly deferred to the agency were “resolved in unpublished dispositions with no precedential value.” Flores Molina v. Garland , 37 F.4th 626, 648 (9th Cir. 2022) (VanDyke, J., dissenting).

Fundamentally, the majority ignores our longstanding practice of presuming that public officers properly discharge Page 56their official duties. Cruz , 146 F.4th at 739. Worse yet, the majority presumes error because the BIA accurately quoted our decision in Benedicto . That obviously cannot be correct. Therefore, I respectfully dissent.