Compassionate Release Cannot Be Used to Collaterally Attack a Federal Conviction; Such Challenges Must Proceed Under 28 U. S. C. §2255

Fernandez v. United States, 608 U. S. ___ (2026) (Barrett, J.) — Decided May 28, 2026

New rule/precedent: A prisoner who collaterally attacks the validity of a federal conviction must proceed through 28 U. S. C. §2255, not the compassionate-release statute, 18 U. S. C. §3582(c)(1)(A)(i). The alleged invalidity of a conviction is not an “extraordinary and compelling” reason for compassionate release.

1. Introduction

Fernandez presented the Supreme Court with a recurring post–First Step Act question: whether the compassionate-release provision (18 U. S. C. §3582(c)(1)(A)(i)) can be used as a vehicle to revisit the correctness of a conviction when traditional postconviction routes (especially 28 U. S. C. §2255) are procedurally constrained or already exhausted. Joe Fernandez, convicted of murder-for-hire and originally sentenced to consecutive life terms, sought compassionate release principally on the ground that he was innocent and that the trial evidence—especially the testimony of cooperating witness Patrick Darge—was unreliable.

The District Court granted relief based on “disquiet” about the verdict; the Second Circuit reversed, holding that conviction-validity challenges are not cognizable as “extraordinary and compelling reasons” under §3582. The Supreme Court granted certiorari to resolve a circuit split and affirmed the Second Circuit.

Parties
  • Petitioner: Joe Fernandez
  • Respondent: United States
Key statutory provisions
  • 18 U. S. C. §3582(c)(1)(A)(i) (compassionate release; “extraordinary and compelling reasons”)
  • 28 U. S. C. §2255 (collateral attack on federal conviction/sentence)
  • 18 U. S. C. §1958 (murder for hire) and 18 U. S. C. §924(j) (firearms offense) (charges/background)

2. Summary of the Opinion

Justice Barrett’s opinion holds that §3582 compassionate release is not a substitute for §2255. The Court reasons that Congress created a “reticulated” and tightly constrained framework for collateral review in §2255 (time limits, limits on successive motions, default rules, and restrictions on relitigation). Allowing prisoners to repackage conviction attacks as compassionate-release motions would permit circumvention of that framework. Accordingly, the Court construes “extraordinary and compelling reasons” to exclude purported conviction invalidity.

The Court further rejects an “actual innocence” carveout, emphasizing that an innocence claim still “challenge[s] the validity of his conviction” and lies “close to the core of habeas corpus.” The judgment below is affirmed.

3. Analysis

3.1 Precedents Cited

The Court’s reasoning is driven less by traditional compassionate-release caselaw than by habeas “channeling” precedent and postconviction structure.

  • Preiser v. Rodriguez, 411 U. S. 475 (1973).
    Influence: The Court treats Preiser as establishing an anticircumvention principle: when a claim is “close to the core of habeas corpus,” prisoners must use the “specific federal habeas corpus statute,” even if another statute’s “literal terms” might be broad enough to cover the claim. The majority analogizes a conviction-validity argument under §3582 to the Preiser attempt to use §1983 to obtain relief that functionally attacks confinement duration.
  • District Attorney's Office for Third Judicial Dist. v. Osborne, 557 U. S. 52 (2009).
    Influence: Quoted for the proposition that other statutes “must be read in harmony” with the habeas framework. The majority uses Osborne to justify reading §3582 against the background of §2255 rather than in isolation.
  • Gonzalez v. Crosby, 545 U. S. 524 (2005).
    Influence: Central analogy: Rule 60(b) could not be used to relitigate the merits of habeas denials because it would “circumvent” strict habeas limits. The Court treats §3582 similarly: a broad-sounding “extraordinary and compelling” standard cannot be used to evade §2255’s successive-motion, default, and finality constraints.
  • Kaufman v. United States, 394 U. S. 217 (1969) and Reed v. Farley, 512 U. S. 339 (1994) (Scalia, J., concurring in part and concurring in judgment).
    Influence: Cited for the principle that claims raised and rejected on direct review generally cannot be relitigated via §2255, underscoring Congress’s preference for finality and structured collateral review.
  • Bousley v. United States, 523 U. S. 614 (1998).
    Influence: Supplies the “cause and prejudice” or “actual innocence” gateway for procedurally defaulted claims, illustrating how Congress and the Court already police the boundary between finality and error-correction within habeas.
  • Herrera v. Collins, 506 U. S. 390 (1993), and Schlup v. Delo, 513 U. S. 298 (1995).
    Influence: Herrera is used to define actual innocence as a challenge to conviction validity (“challenge[s] the validity of his conviction”)—thus bringing innocence within the habeas “core.” Schlup appears because Fernandez invoked it as a habeas-related innocence framework, reinforcing to the majority that his §3582 motion was functionally a collateral attack.
  • Jones v. Hendrix, 599 U. S. 465 (2023).
    Influence: Cited to reinforce the Court’s modern emphasis on the exclusivity and rigidity of congressionally designed postconviction pathways.
  • United States v. Davis, 588 U. S. 445 (2019).
    Influence: Procedural backdrop: Fernandez obtained §2255 relief as to the firearms count after Davis. The majority uses this history to show that §2255 can provide meaningful relief but within strict limits.
  • Brady v. Maryland, 373 U. S. 83 (1963), and Strickland v. Washington, 466 U. S. 668 (1984).
    Influence: The Court references these doctrines to rebut Fernandez’s claim that §3582 should be available when §2255’s substantive standards are “intricate.” The majority treats that argument as an admission that §3582 is being used to dilute §2255’s requirements.
  • Rutherford v. United States, ___ U. S. ___ (2026).
    Influence: Decided the same day; the majority invokes Rutherford to reject the broad claim that “extraordinary and compelling” permits “virtually any reason whatsoever,” and it cites Rutherford to criticize the Sentencing Commission’s “Unusually Long Sentence[s]” category as beyond the Commission’s authority.
  • Circuit-split cases (cited in footnote): United States v. Ferguson; United States v. Escajeda; United States v. West; United States v. Von Vader; United States v. Crandall; United States v. Wesley; United States v. Jenkins (majority-aligned circuits), contrasted with United States v. Trenkler and United States v. Roper (minority view).
    Influence: These cases contextualize the conflict; the Supreme Court’s decision resolves it in favor of the “no conviction attack via compassionate release” rule.

3.2 Legal Reasoning

(1) Structural channeling: §2255 as the exclusive route for conviction validity attacks.

The opinion’s core move is structural: §2255 is the congressionally designed mechanism for collateral attacks on federal convictions, and it carries “tight procedural constraints” (limitations period, successive-motion restrictions, default and relitigation rules). If §3582 could entertain the same attack under the label “extraordinary and compelling,” those constraints would become optional. The Court reads Preiser and Gonzalez to forbid that kind of evasion.

(2) Text, context, and institutional design of compassionate release.

Although the Court acknowledges that Congress did not define “extraordinary and compelling,” it rejects the notion that the phrase is an “empty vessel.” It places special weight on “compelling” as a context-dependent justification for mercy, not a general-purpose vehicle for error correction. It then uses contextual features of §3582 to cabin the phrase: statutory attention to terminal illness, age-based release, rehabilitation as a consideration (but not alone), and the Bureau of Prisons’ designed role as initial evaluator—an institution with expertise in prison conditions and prisoner circumstances, not trial-record error review.

(3) Remedy mismatch as evidence of statutory purpose.

The Court highlights that if a conviction is invalid, the fitting remedy is vacatur—the hallmark of §2255 relief—not “a little less prison time.” That mismatch supports the conclusion that Congress did not mean §3582 to serve as a conviction-correction mechanism.

(4) Application to Fernandez: his motion was, in substance, a conviction attack.

The Court emphasizes the content of Fernandez’s §3582 motion: he argued he “would never commit” the crime, urged the court to distrust trial testimony, and alleged “highly exculpatory” nondisclosures—classic trial-level defects that imply a flawed conviction. The District Court’s rationale—“doubt that the jury’s verdict was correct”—confirmed that the relief was granted as a second look at guilt, not as mercy based on new personal circumstances.

(5) No actual-innocence exception.

Even if federal habeas law’s treatment of freestanding innocence is unsettled (Herrera), the majority holds that an innocence claim still “challenge[s] the validity of [the] conviction” and thus belongs in habeas territory, not §3582. The Court also notes that no court found Fernandez actually innocent; the District Court relied on “disquiet,” not a definitive exoneration.

3.3 The Separate Writings: Competing Visions of §3582

Sotomayor, J., concurring in the judgment (joined by Kagan, J.)

The concurrence rejects the majority’s habeas-channeling limitation as “atextual” and “likely unworkable,” but would affirm on a narrower ground: compassionate release cannot rest “solely on facts a court already considered in imposing the initial sentence,” absent “changed circumstances that developed after sentencing.” In her view, Fernandez simply recycled long-litigated trial arguments; thus, affirmance was available without creating a categorical “collateral attack” bar.

Jackson, J., dissenting

The dissent views §3582(c)(1)(A) as a “safety valve” designed to preserve judicial discretion to mitigate injustice in a determinate sentencing system. It criticizes the majority for importing habeas doctrine into a statute whose text contains no such limitation and for creating line-drawing problems about what counts as an impermissible “collateral attack.” The dissent also stresses the differences between §2255 (mandatory vacatur upon unlawfulness) and §3582 (discretionary mercy without disturbing the conviction’s finality).

3.4 Impact

  • Resolves the circuit split: Conviction- and trial-validity challenges (including innocence-framed arguments) cannot be reintroduced through §3582 as “extraordinary and compelling reasons.”
  • Reinforces §2255 exclusivity and finality: The decision strengthens the postconviction “channeling” model by treating §3582 as non-overlapping with habeas for conviction-validity issues, preventing repeated, time-unlimited relitigation through compassionate release.
  • Constrains district courts’ “second look” use of compassionate release: District court discomfort with a verdict—without new personal circumstances—cannot be transformed into compassionate release if it is essentially doubt about guilt.
  • Shifts strategic litigation: Prisoners will be pushed back toward §2255 (or executive clemency) for innocence and trial-error claims, even where §2255’s limitations period, successive-motion gatekeeping, or default rules make relief difficult.
  • Future boundary disputes: Despite the majority’s confidence, litigation will likely arise over mixed-motive motions (e.g., serious illness plus claimed innocence; mental health claims intertwined with trial competency allegations) and over how to classify arguments as “personal circumstances” versus “conviction attacks.”

4. Complex Concepts Simplified

  • Compassionate release (18 U. S. C. §3582(c)(1)(A)): A mechanism allowing a sentencing court to reduce an otherwise final prison sentence when “extraordinary and compelling reasons” warrant it, typically tied to humanitarian or individualized circumstances (e.g., terminal illness).
  • Collateral attack / habeas substitute: A postconviction attempt to invalidate (or functionally undermine) a conviction or sentence after direct appeal is over—typically pursued through 28 U. S. C. §2255 for federal convictions.
  • §2255’s “procedural constraints”: Rules that limit when and how often a prisoner can seek postconviction relief: a one-year time bar, strict limits on second or successive motions, and default/relitigation doctrines.
  • “Channeling” principle (from Preiser/Gonzalez): If Congress created a specific pathway with strict rules for a type of claim (here, conviction validity via §2255), courts should not interpret other broadly worded provisions to allow end-runs around that pathway.
  • Actual innocence: A claim that the defendant did not commit the crime. The Court treats such a claim as inherently questioning the conviction’s validity, even if framed as a request only to shorten the sentence.

5. Conclusion

Fernandez v. United States draws a bright line: compassionate release is a mercy-based, circumstance-driven sentencing reduction mechanism, not an alternative forum for relitigating guilt or trial error. By anchoring its holding in the anticircumvention logic of Preiser v. Rodriguez and Gonzalez v. Crosby, the Court harmonizes §3582 with Congress’s tightly constrained collateral-review architecture in §2255 and forecloses the use of “extraordinary and compelling reasons” to bypass habeas limits. The decision will likely reduce district courts’ ability to use compassionate release as a broad “second look” tool when the motivating concern is the conviction’s correctness, while redirecting such claims—however compelling on their facts—back into the narrower channels of federal postconviction law.