Rutherford v. United States - Annotated Opinion

 Daniel Rutherford, petitioner in one of these consolidated cases, committed two separate robberies at gunpoint in 2003. A jury convicted him of several crimes, including two counts of using and carrying a firearm during a crime of violence, in violation of §924(c). At the time, a defendant convicted of two §924(c) counts was subject to a mandatory 25-year sentence of imprisonment for the second offense that would be “stacked” upon—that is, run consecutively to—the first offense’s mandatory penalty (which, for Rutherford, was 7 years). See Deal v. United States, 508 U. S. 129, 132–137 (1993). Rutherford’s two §924(c) counts thus earned him a 32-year minimum sentence, resulting in a total sentence of over 42 years of imprisonment.

 Johnnie Carter, petitioner in the other case, received an even longer mandatory minimum under §924(c). In 2007, Carter participated in a string of armed bank robberies. Among other offenses, he was convicted of three §924(c) counts. Carter was sentenced to 70 years of imprisonment—57 of which came from his stacked §924(c) convictions.

 Years after Rutherford and Carter were each sentenced, Congress passed the First Step Act of 2018, 132 Stat. 5194, “a landmark piece of legislation that changed the federal criminal-sentencing system in numerous respects,” Hewitt v. United States, 606 U. S. 419, 424 (2025). Among the changes was the elimination of §924(c)’s 25-year stacking requirement for first-time offenders: A 25-year consecutive sentence is now required only for a “ ‘violation . . . that occurs after a prior conviction under this subsection has become final.’ ” §403(a), 132 Stat. 5222. This change would have made a significant difference for both Rutherford and Carter. Had either been sentenced today, his mandatory minimum under §924(c) would have been much shorter: 14 years for Rutherford and 21 years for Carter.

  Obviously, Rutherford and Carter—not to mention the many other similarly situated prisoners—would like the benefit of the Act’s amendment to §924(c). As a rule, however, changes to sentencing statutes apply only to offenses committed after the statute’s effective date. See Hewitt, 606 U. S., at 424. The Act departs from this rule in a very limited respect: Its more lenient penalties apply to offenses committed before the statute’s enactment “if a sentence for the offense has not been imposed as of such date of enactment.” §403(b), 132 Stat. 5222. But for defendants sentenced before the Act, the general rule of nonretroactivity remains in place. See Hewitt, 606 U. S., at 427.

 Rutherford and Carter acknowledge that they do not qualify for the reduced penalties. Nonetheless, each has invoked the Act as a ground for compassionate release under §3582(c)(1)(A)(i).

 At this point, some background on the compassionate release provision is helpful. This provision, which was part of the Sentencing Reform Act of 1984, allows a district court to reduce a prison term “after considering the factors set forth in section 3553(a) to the extent that they are applicable” if the court finds that “extraordinary and compelling reasons warrant such a reduction” and that “such a reduction is consistent with applicable policy statements issued by the Sentencing Commission.” 18 U. S. C. §3582(c)(1)(A). Congress did not specify what “extraordinary and compelling” reasons might warrant compassionate release. It did, however, authorize the Commission to promulgate policy statements that “describe what should be considered extraordinary and compelling reasons for sentence reduction, including the criteria to be applied and a list of specific examples.” 28 U. S. C. §994(t). The Commission first issued an applicable policy statement in 2006 and amended it the following year to identify several “ ‘extraordinary and  compelling’ ” reasons for compassionate release: “ ‘terminal illness’ ”; “ ‘permanent physical or medical condition’ ”; “ ‘deteriorating physical or mental health because of the aging process’ ”; “ ‘death or incapacitation of the defendant’s only family member capable of caring for the defendant’s minor child or minor children’ ”; and a catchall category for “ ‘other’ ” reasons not specifically enumerated, “ ‘[a]s determined by the Director of the Bureau of Prisons.’ ” United States Sentencing Commission, Guidelines Manual Supp. to App. C, Amdt. 698 (Nov. 2011) (USSG). In 2016, the Commission streamlined the statement to recognize four categories of reasons: “ ‘Medical Condition of the Defendant,’ ” “ ‘Age of the Defendant,’ ” “ ‘Family Circumstances,’ ” and “ ‘Other Reasons.’ ” Id., Amdt. 799.

 When Congress passed the First Step Act in 2018, it amended the compassionate release provision. While it left the “extraordinary and compelling” standard untouched, it opened a new procedural route for seeking a reduction. Originally, only the Bureau of Prisons could ask a district court to reduce a prisoner’s sentence. The Act gives the prisoner greater agency: He may request that the Bureau file a motion on his behalf, and, if it declines or fails to do so within 30 days, he may file a motion himself. 18 U. S. C. §3582(c)(1)(A). The Act also obligates the Bureau to take certain steps when a prisoner is diagnosed with a terminal illness or is mentally or physically unable to file his own motion. §§3582(d)(2)(A)–(B). For example, the Bureau must inform “the defendant’s attorney, partner, and family members” about the defendant’s condition and that they can file a sentence-reduction request on his behalf. Ibid.

 The Commission lost a quorum shortly after the Act went into effect and did not update its policy statement to address these changes. Most Circuits held that the existing statement—which mentioned only motions filed by the Bureau—did not apply to motions filed by prisoners. See United States v. Andrews, 12 F. 4th 255, 259 (CA3 2021)  (collecting cases); but see United States v. Bryant, 996 F. 3d 1243, 1247–1248 (CA11 2021) (applying the policy statement). So for this new class of motions, most courts interpreted the statute without input from the Commission.

 They were soon forced into uncharted territory. Relevant here, courts faced a surge of motions from prisoners who invoked the Act’s nonretroactive sentencing amendments as grounds for compassionate release. Most Courts of Appeals held that a nonretroactive amendment to a mandatory minimum cannot serve as an “extraordinary and compelling” reason for a reduced sentence.² A minority, however, concluded that the disparity produced by such an amendment can justify a reduced sentence—at least when considered in combination with other factors.³

 The Commission eventually regained a quorum, entered the fray, and sided with the minority. In 2023, it amended its policy statement to add a new category to the list of “extraordinary and compelling” reasons: “Unusually Long Sentence.”⁴ See USSG App. C, Amdt. 814. Under this category:

“ ‘If a defendant received an unusually long sentence and has served at least 10 years of the term of imprisonment, a change in the law (other than an amendment to the Guidelines Manual that has not been made retroactive) may be considered in determining whether  the defendant presents an extraordinary and compelling reason, but only where such change would produce a gross disparity between the sentence being served and the sentence likely to be imposed at the time the motion is filed, and after full consideration of the defendant’s individualized circumstances.’ ” Ibid.

Three Commissioners voted against the amendment and accused the Commission of making “a seismic structural change to our criminal justice system without congressional authorization or directive.” United States Sentencing Commission, Public Meeting Tr. 60 (Apr. 5, 2023).

 Back to the cases at hand. After the First Step Act became effective, Rutherford moved for compassionate release. He argued that “extraordinary and compelling” reasons warrant a reduction because he had received an “unusually long sentence under a statutory provision that Congress has since found too punitive but has not made retroactively applicable.” 2 App. in No. 23–1904 (CA3), p. 58. In addition, he noted that he suffered from obesity and hypertension, which rendered him more vulnerable to the COVID–19 virus.

 While Rutherford’s motion was pending, the Third Circuit held that the nonretroactive change to §924(c) cannot serve as an “extraordinary and compelling” reason for compassionate release. See Andrews, 12 F. 4th, at 260–262. The District Court then denied Rutherford’s motion, and Rutherford appealed that ruling to the Third Circuit. While his appeal was pending, the Commission amended its policy statement, and Rutherford asked the Court of Appeals to consider the effect of this statement in the first instance.

 The Third Circuit affirmed the District Court’s denial of Rutherford’s motion. See 120 F. 4th 360, 363 (2024). The court acknowledged that Congress had authorized the Commission to issue a policy statement fleshing out the  meaning of “ ‘extraordinary and compelling’ ” reasons under §3582(c)(1)(A)(i). Id., at 375. But that statement, the court said, cannot “go beyond what Congress intended.” Id., at 376. And as the Third Circuit had already explained in Andrews, it makes no sense to “ ‘construe Congress’s nonretroactivity directive as simultaneously creating an extraordinary and compelling reason for early release.’ ” 120 F. 4th, at 376 (quoting Andrews, 12 F. 4th, at 261). Because the Commission’s amended policy statement “conflicts with the will of Congress,” the court held, it “cannot be considered in determining a prisoner’s eligibility for compassionate release.” 120 F. 4th, at 376.

 Carter’s case is similar. After the Commission amended its policy statement, he moved for a reduction under §3582(c)(1)(A)(i), arguing that his sentence is “unusually long” and that the Act created a “ ‘gross disparity’ ” between his original sentence and the one he would receive under the amended §924(c). 2 App. in No. 24–1115 (CA3), pp. 206–207. He also pointed to his strong family ties, rehabilitation, and good conduct while incarcerated as additional reasons for a reduction. Bound by Andrews, the District Court denied Carter’s motion, and Carter appealed. While the appeal was pending, the Third Circuit decided Rutherford’s case. Shortly thereafter, it summarily affirmed the denial of Carter’s motion. See 2024 WL 5339852, *1 (Dec. 2, 2024).

 The Circuits remain divided over whether the disparity created by a nonretroactive change to sentencing law, like the Act, can constitute an “extraordinary and compelling reaso[n]” that “warrant[s]” compassionate release.⁵ 18  U. S. C. §3582(c)(1)(A)(i). We granted certiorari in these consolidated cases to resolve the split. 605 U. S. 1001 (2025).

 The compassionate release provision empowers a district court to “reduce the term of imprisonment” if it finds, among other things, that “extraordinary and compelling reasons warrant such a reduction.” Ibid. The term “extraordinary” means “most unusual,” “far from common,” or “having little or no precedent.” Webster’s Third New International Dictionary 807 (1976); see 5 Oxford English Dictionary 614 (2d ed. 1989) (“[o]ut of the usual or regular course or order”). “Compelling,” meanwhile, means “tending to convince or convert by or as if by forcefulness of evidence.” Webster’s Third New International Dictionary, at 463; see 3 Oxford English Dictionary, at 600 (“irresistible; demanding attention, respect”). Putting these definitions together, “extraordinary and compelling” reasons for  compassionate release are those that are especially unusual and convincing.

 The disparity created by Congress’s amendment to §924(c) does not satisfy this definition. Far from “extraordinary,” nonretroactive amendments to criminal penalties are the norm. “Ordinarily,” we have explained, “statutory changes to federal penalties only benefit future offenders.” Hewitt, 606 U. S., at 424; see 1 U. S. C. §109. Congress sometimes deviates from the default, as it did in the Act, by extending the revised penalties to defendants not yet sentenced. Even then, some measure of prospectivity is par for the course: The “ordinary practice” is to “withhol[d] that change from defendants already sentenced.” Dorsey v. United States, 567 U. S. 260, 280 (2012). And regardless of whether the line is drawn at the offense or the sentence, a nonretroactive penalty change necessarily creates a disparity between those who receive the benefit and those who do not. See ibid. (“[D]isparities . . . reflec[t] a line-drawing effort” and “will exist whenever Congress enacts a new law changing sentences”). Such a disparity is an unexceptional feature of a system in which nonretroactivity is the default.

 As for “compelling”: It is hard to see how Congress’s deliberate decision not to extend newly reduced penalties to those already sentenced could be a convincing reason that “warrant[s]” a reduction. 18 U. S. C. §3582(c)(1)(A)(i). As we have explained, Congress decided to apply the lower prison sentences to some violators of §924(c)—those for whom a sentence had not yet been imposed—but not others. §403(b), 132 Stat. 5222. In doing so, Congress “reinforced its interest in finality and avoided burdening district courts with additional litigation.” Hewitt, 606 U. S., at 437–438 (plurality opinion). Treating the disparity resulting from §924(c)’s amendment as a compelling reason for reducing a sentence would undermine Congress’s choice to leave the sentence intact.

  It would also fall well outside the heartland of compassionate release, which has long been defined by a prisoner’s personal circumstances. After Congress enacted §3582(c)(1)(A)(i), the Bureau of Prisons explained that the standards for compassionate release mirrored those of a prior sentence-reduction statute and that “[r]eleases have been most often applied in cases where the inmate is terminally ill.” 59 Fed. Reg. 1238 (1994); see 28 CFR §572.40(a) (1984) (providing that the Bureau would move for a sentence reduction under 18 U. S. C. §4205(g) (1982 ed.) “in particularly meritorious or unusual circumstances,” such as “if there is an extraordinary change in an inmate’s personal or family situation or if an inmate becomes severely ill”). The Act’s heightened procedural requirements are similarly tied to personal circumstances: The Bureau must provide extra assistance to a movant diagnosed with a terminal illness or who is physically or mentally unable to submit a request on his own. §3582(d)(2) (2018 ed.). And from 2007 until 2023, the Sentencing Commission highlighted essentially three grounds for compassionate release: (1) medical condition, (2) age, and (3) family circumstances.⁶ See USSG §1B1.13, comment., n. 1 (Nov. 2021).

 We need not decide whether there are reasons beyond personal circumstances that could qualify as “extraordinary and compelling.” Until very recently, no one thought that  nonretroactive sentencing amendments were among them.⁷ They are not.

 Petitioners, echoed by the dissent, insist that our reading artificially limits the text. The terms “extraordinary and compelling,” they say, are “ ‘comprehensive and flexible,’ ” requiring a “ ‘case-by-case’ inquiry into the ‘totality of the circumstances.’ ” Brief for Petitioner Rutherford 15, 17; see Brief for Petitioner Carter 41. And under this “ ‘flexible, all-things-considered approach,’ ” a judge may consider the disparity resulting from Congress’s amendment to §924(c). Brief for Petitioner Rutherford 17; see post, at 7–8 (opinion of Sotomayor, J.).

 While the terms “extraordinary” and “compelling” leave room for judgment, they are not so flexible as to encompass any consideration. Their meaning depends on context: A reason is “extraordinary” and “compelling” only if it is sufficiently unusual and convincing to “warrant” compassionate release. 18 U. S. C. §3582(c)(1)(A)(i); see Fernandez v. United States, ___ U. S. ___, ___, and n. 4 (2026) (slip op., at 10, and n. 4). As we have explained, the disparity resulting from Congress’s amendment to §924(c) tracks ordinary sentencing practice and reflects Congress’s deliberate choice to extend relief to some prisoners and not others. Thus, the disparity is neither an “extraordinary” nor a “compelling” reason that warrants a reduction in sentence.

  Both petitioners and the dissent note that when Congress empowered the Sentencing Commission to describe the grounds for compassionate release, it included only one express limitation: “Rehabilitation of the defendant alone shall not be considered an extraordinary and compelling reason.” 28 U. S. C. §994(t). By ruling out just one reason, they argue, Congress impliedly authorized the district court to consider all other relevant information. Post, at 11–12.

 This argument fares no better. We have repeatedly emphasized that “ ‘[t]he force of any negative implication . . . depends on context.’ ” NLRB v. SW General, Inc., 580 U. S. 288, 302 (2017) (quoting Marx v. General Revenue Corp., 568 U. S. 371, 381 (2013)); see also A. Scalia & B. Garner, Reading Law: The Interpretation of Legal Texts 107–111 (2012). A speaker’s choice to rule out one item does not always mean that the rest of the universe is on the table. Imagine that a dietitian cautions her patient, “Pasta alone is not an acceptable dinner.” Though the patient might fervently wish otherwise, this advice does not license any meal besides standalone pasta. (Fried chicken with a side of funnel cakes? A bowl of Lucky Charms?) Instead, the dietitian surely had some reason—like the patient’s past eating habits—to caution against this specific choice.

 So too with §994(t). By addressing rehabilitation, Congress did not impliedly bless all other considerations. Instead, it had a particular reason—namely, past sentencing practice—to single out this factor. Before the Sentencing Reform Act, “[b]oth indeterminate sentencing and parole were based on concepts of the offender’s possible, indeed probable, rehabilitation, a view that it was realistic to attempt to rehabilitate the inmate and thereby to minimize the risk that he would resume criminal activity upon his return to society.” Mistretta v. United States, 488 U. S. 361, 363 (1989). Guided by this aim, the judge and parole officer each made decisions based on “their own assessments of the offender’s amenability to rehabilitation.” Ibid. When the  Sentencing Reform Act eliminated parole and instituted a new sentencing regime, it “reject[ed] imprisonment as a means of promoting rehabilitation.” Id., at 367; see 18 U. S. C. §3582(a) (“[I]mprisonment is not an appropriate means of promoting correction and rehabilitation”). Ruling out “rehabilitation alone” as a basis for compassionate release therefore underscores the break from the old system. It suggests nothing about what other reasons might qualify as “extraordinary and compelling.”

 And in petitioners’ telling, the list is long. They emphasize that when imposing or modifying a sentence, a district court ordinarily enjoys “broad discretion to consider all relevant information” unless prohibited by the Constitution or by Congress. Concepcion v. United States, 597 U. S. 481, 491 (2022); see 18 U. S. C. §3661 (“No limitation shall be placed on the information concerning the background, character, and conduct of a person . . . which a court . . . may receive and consider for the purpose of imposing an appropriate sentence”). Section 3582(c)(1)(A) expressly recognizes that discretion, they argue, by providing that a court may grant relief “after considering the factors set forth in section 3553(a) to the extent that they are applicable.” Those factors include “the kinds of sentences available,” “the kinds of sentence and the sentencing range established for . . . the applicable category of offense,” and “the need to avoid unwarranted sentence disparities.” §§3553(a)(3)–(4), (6). A factor like the disparity caused by a nonretroactive sentencing amendment, petitioners contend, fits right in. See also post, at 10–11, 16–17.

 Petitioners skip a step. Before determining the extent of a reduction based on the §3553(a) factors, a court must first ensure that a movant is part of the “limited class of prisoners” who are “eligibl[e]” for such a reduction. Dillon v. United States, 560 U. S. 817, 825, 827 (2010). And whether a prisoner is eligible depends on whether he can offer “extraordinary and compelling” reasons that “warrant”  compassionate release, not on the §3553(a) factors. §3582(c)(1)(A)(i); see id., at 827. This gatekeeping requirement is not, as petitioners seem to believe, a free-for-all. It is a distinct analytical step that imposes independent and ascertainable limits on access to compassionate release.

 Concepcion v. United States, 597 U. S. 481, is not to the contrary. Concepcion involved sentence-modification proceedings under a different provision of the Act. See id., at 486–487. All agreed that the defendant was eligible for a reduction under that provision; the only question was what type of information the District Court could consider in modifying his sentence. See id., at 488 (explaining that “[t]he Government conceded [the defendant’s] eligibility for relief ”). Because that provision lacks any limiting language, we concluded that a court could consider changes to the Guidelines, as well as intervening facts, when calculating a new sentence under it. Id., at 500. Today’s cases differ from Concepcion because they concern whether a prisoner is eligible for compassionate release in the first place. And on that score, Congress has “expressly cabined district courts’ discretion,” id., at 495, by prohibiting a reduction in sentence unless a court finds that “extraordinary and compelling” reasons warrant it.

 Finally, petitioners downplay the conflict between the Act and their interpretation of §3582(c)(1)(A)(i). True, Congress chose not to reduce §924(c) penalties on a categorical basis for already sentenced defendants. But petitioners’ approach is not categorical: The disparity between a pre- and post-Act sentence would serve as just one of at least two considerations that, taken together, warrant compassionate release. And reducing a sentence based on an all-things-considered judgment, petitioners argue, does not contradict Congress’s decision to leave a class of offenders ineligible for a sentencing reduction. See also post, at 15–16.

  The implications of this argument are sweeping. Indeed, its logic extends well beyond nonretroactive sentencing amendments to disagreement with the length of any punishment on the books. Take a judge who believes that a mandatory minimum is unduly harsh. Could she treat the severity of the sentence as an “extraordinary and compelling reason” for compassionate release? Petitioners refused to rule out this possibility at oral argument. Tr. of Oral Arg. 11–15, 44–49. Yet “[i]t is the legislature, not the Court, which is to define a crime, and ordain its punishment.” United States v. Wiltberger, 5 Wheat. 76, 95 (1820). Treating the severity of a mandatory penalty as a reason for compassionate release rejects Congress’s judgment that the punishment fits the crime.⁸ And petitioners’ proposal—injecting other factors into the mix of extraordinary and compelling reasons—does not solve the problem. The Sixth Circuit put it well: “[A]dding a legally impermissible ground to three insufficient factual considerations does not entitle a defendant to a sentence reduction.” United States v. Jarvis, 999 F. 3d 442, 444 (2021).

 This brings us to the Sentencing Commission. Recall that Congress authorized the Commission to promulgate policy statements that “describe what should be considered extraordinary and compelling reasons for sentence reduction,” 28 U. S. C. §994(t), and provided that any grant of  compassionate release must be consistent with the Commission’s policy statements, 18 U. S. C. §3582(c)(1)(A). Recall, too, that the Commission’s current policy statement allows consideration of nonretroactive legal changes in certain circumstances. Petitioners contend that our interpretation of the statute must yield to the policy statement.

 Petitioners misunderstand the Commission’s role. Congress has empowered the Commission to “give meaning” to the compassionate release provision by identifying the circumstances that constitute “extraordinary and compelling” reasons for release. Loper Bright Enterprises v. Raimondo, 603 U. S. 369, 394 (2024). Even so, the Commission’s policy statements must be “consistent with” the governing statute, 28 U. S. C. §994(a), and courts have a duty to “independently interpret the statute and effectuate the will of Congress,” 603 U. S., at 395. We are not bound to follow the Commission’s guidance when it “exceed[s its] statutory authority” by adopting a definition of a term that is inconsistent with the statute. Batterton v. Francis, 432 U. S. 416, 426 (1977). Put differently, the Commission’s interpretation must land within the statutory goalposts, no matter how wide they might be. Cf. Koons v. United States, 584 U. S. 700, 707 (2018) (“[P]olicy statements cannot make a defendant eligible when [the statute] makes him ineligible”).

 We need not nail down the precise boundaries of the phrase “extraordinary and compelling” to conclude that “with regard to the particular dispute in [this] case,” the statute’s language “has a plain and unambiguous meaning.” Robinson v. Shell Oil Co., 519 U. S. 337, 340 (1997). The statutory text and structure make clear that Congress’s nonretroactive change to §924(c)—considered by itself or in combination with other factors—cannot make a prisoner eligible for compassionate release. To the extent that it counsels otherwise, the Commission’s policy statement is invalid.

 Compassionate release is available only when a court finds that “extraordinary and compelling reasons warrant” a sentence reduction. 18 U. S. C. §3582(c)(1)(A)(i). The disparity that results from Congress’s decision to leave a sentence untouched cannot serve as one of those reasons. The judgments of the Third Circuit are affirmed.

 In 2003, when he was 25 years old, petitioner Daniel Rutherford committed two armed robberies in which no one was hurt and during which he took a watch and roughly $1,300 worth of other jewelry and cash. He was convicted of two violations of 18 U. S. C. §924(c)(1). At the time, those violations together carried a mandatory minimum sentence of 32 years, to run consecutive to his sentence for the robberies. In total, Rutherford was sentenced to 42.5 years in prison. As two judges noted in Rutherford’s direct appeal of his conviction, that sentence “would be unthinkable in many state systems.” 236 Fed. Appx. 835, 845 (CA3 2007) (Ambro, J., joined by McKee, J., concurring).

 In 2007, petitioner Johnnie Markel Carter and others robbed several banks and stole about $250,000. As with Rutherford’s robberies, no one fired a gun and no one was hurt. Carter, who opted to go to trial, was convicted of three violations of §924(c) and later sentenced to a term of 70 years in prison, 57 of which were from mandatory minimums for the §924(c) violations. (Carter’s co-conspirators, by contrast, took plea deals and received sentences ranging from 10 to 23 years.) For Carter, who was 29 years old at the time, the 70-year sentence was a de facto life sentence.

 In the two decades since their convictions, both Rutherford and Carter have turned their lives around. Rutherford has completed over 50 educational courses and received only two minor infractions in the past decade. He also secured a job for after he is released, which would enable him to help support his deceased sister’s five children.

  Carter has, as the District Court put it, “become the kind of model prisoner that our system tries, but too often fails, to produce.” 711 F. Supp. 3d 428, 440 (ED Pa. 2024) (case below). While in prison, Carter obtained a GED, completed vocational training programs, took parenting classes to improve his relationship with his adult children, maintained a perfect disciplinary record for the last eight years, and provided counseling and spiritual guidance to his fellow inmates.

 In 2018, Congress passed the First Step Act and greatly reduced the mandatory minimum for first-time §924(c) offenders like Rutherford and Carter. Had the amendments been in effect when Rutherford was sentenced, he would have faced a 14-year mandatory minimum, less than half of the 32-year mandatory minimum he received. Carter would have been subject to a 21-year mandatory minimum, 36 years shorter than the 57-year mandatory minimum he received. Because Congress did not make its changes retroactive, Rutherford and Carter were not automatically eligible for sentence reductions based on those changes.

 In the wake of the First Step Act, the United States Sentencing Commission in 2023 adopted §1B1.13(b)(6), a policy statement allowing courts, in limited circumstances, to consider sentencing disparities created by nonretroactive changes in law. Specifically, courts may grant relief based on such a change in law only when four requirements are met: (1) the defendant received an “unusually long sentence”; (2) the defendant has served at least 10 years of that sentence; (3) the disparity is “gross”; and (4) the district court has taken “full consideration of the defendant’s individualized circumstances.” USSG §1B1.13(b)(6) (Nov. 2025). Outside of those circumstances, “a change in the law . . . shall not be considered” for purposes of determining  whether an “extraordinary and compelling reason exists” to reduce the sentence. §1B1.13(c).

 Invoking this policy statement, Rutherford and Carter moved for compassionate release under 18 U. S. C. §3582(c)(1)(A), which allows a district court to reduce an individual’s term of imprisonment “if it finds that” “extraordinary and compelling reasons warrant such a reduction” “after considering the factors set forth in section 3553(a) to the extent they are applicable.” Rutherford and Carter argued that there are extraordinary and compelling reasons to reduce their sentences based on a combination of their remarkable personal circumstances as well as the fact that they would have received much shorter sentences under today’s law. The District Courts, however, held that the Commission’s policy statement is invalid and denied petitioners relief, and the Court of Appeals affirmed.

 Before 1984, federal judges had broad discretion when sentencing defendants. Statutes provided minimums and maximums for judges to work between, but within those ranges, judges could determine for themselves what sentence was appropriate. See Mistretta v. United States, 488 U. S. 361, 363–366 (1989). This system allowed judges to consider individual defendants’ circumstances, but it also led to “[s]erious disparities in sentences.” Id., at 365.

 To address those disparities, Congress passed the Sentencing Reform Act of 1984. The Act created the United States Sentencing Commission, which Congress tasked with studying federal sentencing and promulgating the Sentencing Guidelines. 28 U. S. C. §994(a). The Guidelines instruct judges on how to sentence defendants, including how to account for different factual circumstances like the defendant’s background, the seriousness of the defendant’s conduct, and more, within the context of the considerations set forth in 18 U. S. C. §3553(a). See 28 U. S. C. §§991(b), 994(f ), 994(m).¹

 Congress gave the Commission an important policymaking role in the compassionate-release context as well. Under the statute, district courts can grant compassionate release and reduce a defendant’s sentence if they find that there are “extraordinary and compelling reasons” for doing so. 18 U. S. C. §3582(c)(1)(A)(i). Congress neither specified what constitutes “extraordinary and compelling reasons” for relief nor told district courts how to determine if such reasons exist. It entrusted the Commission with those responsibilities instead, instructing district courts that any grant of relief must be “consistent with applicable policy  statements issued by the Sentencing Commission.” §3582(c)(1)(A).

 Congress tasked the Commission with issuing “general policy statements regarding . . . the appropriate use of ” compassionate release. 28 U. S. C. §994(a)(2)(C). Congress also required the Commission to “describe what should be considered extraordinary and compelling reasons for sentence reduction, including the criteria to be applied and a list of specific examples.” §994(t). The only limit Congress imposed was that “[r]ehabilitation of the defendant alone shall not be considered an extraordinary and compelling reason.” Ibid.

 Congress’s express delegation of authority to the Commission limits this Court’s role in these cases. Congress often “authorize[s]” agencies “to exercise a degree of discretion.” Loper Bright Enterprises v. Raimondo, 603 U. S. 369, 394 (2024). “For example, some statutes ‘expressly delegat[e]’ to an agency the authority to give meaning to a particular statutory term.” Ibid. (quoting Batterton v. Francis, 432 U. S. 416, 425 (1977); emphasis deleted). “Others empower an agency to prescribe rules to ‘fill up the details’ of a statutory scheme.” 603 U. S., at 395 (quoting Wayman v. Southard, 10 Wheat. 1, 43 (1825)). Others still authorize agencies to “regulate subject to the limits imposed by a term or phrase that ‘leaves agencies with flexibility,’ . . . such as ‘appropriate’ or ‘reasonable.’ ” 603 U. S., at 395.

 The Sentencing Reform Act does all those things. It instructs the Commission to issue guidance on the “appropriate use of ” compassionate release. §994(a)(2)(C). It also directs the Commission to give meaning to statutory terms and fill up the details of the statutory scheme by “describ[ing] what should be considered extraordinary and compelling reasons for sentence reduction,” identifying “the  criteria to be applied,” and providing “a list of specific examples.” §994(t).

 When a statute expressly delegates authority to an agency in this way, “[a] reviewing court is not free to set aside” the agency’s actions “simply because it would have interpreted the statute in a different manner.” Batterton, 432 U. S., at 425 (citing American Telephone & Telegraph Co. v. United States, 299 U. S. 232, 235–237 (1936)). Instead, the Court must determine “ ‘the boundaries of [the agency’s] delegated authority’ ” and “ensur[e] the agency has engaged in ‘reasoned decisionmaking’ within those boundaries.” Loper Bright, 603 U. S., at 395.

 In short, under the statutory scheme that Congress created, the Commission has the leading role of defining the words “extraordinary and compelling” and thus setting the scope of compassionate release. This Court, by contrast, has the more modest role of ensuring that the Commission does not act unreasonably.

 Start with the terms “extraordinary” and “compelling.” Rather than imposing any bright-line limits on courts’ discretion to consider or weigh information, these terms invite open-ended, fact- and context-specific inquiries into the  totality of the circumstances in individual cases to determine whether they rise to an unusual level. In other words, whether a given set of reasons is “extraordinary and compelling” as applied to a particular defendant’s case is a question of degree, not of kind.

 This Court’s cases interpreting similarly broad terms across different contexts support this understanding. This Court has opined that it is “impossible to anticipate and define every situation that might” qualify as involving “extraordinary circumstances” that warrant federal interference with an ongoing state criminal prosecution. Kugler v. Helfant, 421 U. S. 117, 124–125 (1975). It has also “struggle[d] in vain for any verbal formula that will supply a ready touchstone” for distinguishing between “ordinary” and “extraordinary” business expenses. Welch v. Helvering, 290 U. S. 111, 114–115 (1933).

 Given the difficulty in defining with precision what makes something “unusual,” this Court has held, in the sentencing context, that such determinations fall within district courts’ discretion. In Koon v. United States, 518 U. S. 81 (1996), the Court noted that a district court’s finding that a factor at sentencing was “unusual or exceptional,” such that a departure from the then-binding Sentencing Guidelines was warranted, “embodies the traditional exercise of discretion by a sentencing court.” Id., at 98. That was because such a judgment reflected the district court’s “ ‘special competence’ ” in evaluating “ ‘the fact-specific circumstances of the case’ ” that set it apart from others. Id., at 99. Similarly, when identifying cases that are “ ‘exceptional,’ ” the Court has explained that the inquiry must be carried out by “[d]istrict courts . . . in the case-by-case exercise of their discretion, considering the totality of the circumstances,” as “ ‘[t]here is no precise rule or formula for making these determinations.’ ” Octane Fitness, LLC v. ICON Health & Fitness, Inc., 572 U. S. 545, 554 (2014).

  The Commission’s policy statement fits comfortably within these capacious parameters. To start, the Com- mission requires courts to take “full consideration of  the defendant’s individualized circumstances.” USSG §1B1.13(b)(6). Individuals also must meet three other discrete criteria, including that they are serving an “unusually long” sentence and are facing a “gross” (i.e., unusually large) sentencing disparity. Ibid. Even if all those criteria are met, the Commission’s policy statement does not require courts to grant relief. A court can grant relief only if it concludes that there are extraordinary and compelling reasons after taking in all relevant information about a defendant’s specific circumstances. Ibid.

 The majority’s chief rejoinder is that nonretroactive changes in law are “the norm.” Ante, at 9. True, but so too are many of the other developments that the majority agrees may support compassionate release. Everyone ages. Everyone falls ill. Everyone loses family members and loved ones. These realities are, unfortunately, also “the norm.” Yet all agree that courts may properly consider a defendant’s age, physical health, and family circumstances when determining whether to grant compassionate release. See ante, at 10. That concession reflects the basic reality that facts which are ordinary in isolation can nonetheless combine to form an extraordinary whole. Accordingly, the key question is not whether any one reason for granting relief is extraordinary or compelling on its own or in the abstract, but whether a defendant’s circumstances, taken together, are so extraordinary and compelling as to warrant reducing his sentence. Because answering that question calls for a full assessment of the totality of the circumstances and all relevant information, the words “extraordinary and compelling” do not themselves bar district courts from ever considering nonretroactive changes in law.

 Statistics also undercut the majority’s suggestion that the Commission’s policy statement permits relief in  ordinary cases. Almost no one has received relief under the Commission’s criteria. There are currently over 130,000 individuals serving sentences for federal convictions. United States Sentencing Commission, QuickFacts: Individuals in the Federal Bureau of Prisons (May 2025), https:// www.ussc.gov/sites/default/files/pdf/research-and-publications/ quick-facts/BOP_2025.pdf (archived at https://perma.cc/ K2QS-JCR5). In Fiscal Year 2024, defendants moved for compassionate release 3,015 times nationwide. United States Sentencing Commission, Compassionate Release Data Report, FY 2024, Table 2 (Mar. 2025), https://www. ussc.gov/sites/default/files/pdf/research-and-publications/ federal-sentencing-statistics/compassionate-release/FY24- Compassionate-Release.pdf (archived at https://perma.cc/ FQE2-W57L). District courts granted relief under §1B1.13(b)(6) for sentencing disparities caused by any change in law (not just the changes to 18 U. S. C. §924(c)) just 98 times. Compassionate Release Data Report, Table 10. That small number suggests that §1B1.13(b)(6) has not opened the floodgates to sentence reductions in unexceptional cases.

 Longstanding tradition, which Congress has not disturbed, also supports the Commission’s policy statement.

 For centuries, courts have enjoyed “broad discretion to consider all relevant information at an initial sentencing hearing, consistent with their responsibility to sentence the whole person before them.” Concepcion v. United States, 597 U. S. 481, 491 (2022). “That discretion also carries forward to later proceedings that may modify an original sentence.” Ibid. At these proceedings, district courts can modify initial sentences for a variety of reasons, including following a reversal on appeal or a retroactive change in law. They may also consider whatever evidence they find germane, including “intervening changes of law (such as  changes to the Sentencing Guidelines)” when deciding whether to reduce a sentence. Id., at 486. Indeed, courts may even consider nonretroactive changes to the Sentencing Guidelines. Id., at 499–500.

 In light of this centuries-old tradition, if Congress wishes to restrict what courts may consider at sentence-modification proceedings, it must say so expressly. “The only limitations on a court’s discretion to consider any relevant materials at an initial sentencing or in modifying that sentence are those set forth by Congress in a statute or by the Constitution.” Id., at 494. (All agree that the Constitution does not expressly bar consideration of nonretroactive changes in law.) When Congress is silent, this Court infers that Congress did not intend to limit sentencing courts’ discretion because “Congress is not shy about placing such limits where it deems them appropriate” and has “ ‘shown that it knows how to direct sentencing practices in express terms.’ ” Id., at 494, 497 (quoting Kimbrough v. United States, 552 U. S. 85, 103 (2007)).

 A compassionate-release proceeding, in which a court decides whether to reduce a defendant’s sentence (and if so, by how much), is obviously a sentence-modification proceeding. Changes in law are also plainly relevant to a court’s determination: The fact that a defendant would be eligible for a significantly lower sentence for the same conduct today due to a change in law undoubtedly could inform a court’s decision as to whether it should reduce his sentence. Allowing courts to consider this information, then, is consistent with courts’ traditional discretion.

 Congress’s words here do not limit courts’ discretion as to what they may consider. As explained, the words “extraordinary and compelling” do not foreclose the approach the Commission adopted in §1B1.13(b)(6). Congress’s delegation to the Commission, moreover, contained just one express restriction: “Rehabilitation of the defendant alone shall not be considered an extraordinary and compelling  reason.” 28 U. S. C. §994(t). That restriction, however, does not bar courts from considering rehabilitation in conjunction with other facts and evidence. Nor does it bar the Commission from allowing courts to consider evidence of rehabilitation alongside other facts, as the Commission has done. See USSG §1B1.13(d). That specific restriction thus poses no limit on what evidence courts may consider, and in fact confirms that Congress knows how to limit compassionate-release practices explicitly when it wants to.

 The majority tries to downplay Congress’s treatment of rehabilitation by analogizing to a conversation with a dietitian. Ante, at 12. An instruction from a dietitian not to eat pasta alone, the majority says, is not a “license” to eat unhealthy food, so Congress’s express limit on rehabilitation, the majority reasons, “did not impliedly bless all other considerations” either. Ibid. This analogy misses the mark. Unlike a judge who enters a sentence-modification proceeding assuming she may consult all relevant information not expressly excluded by law, see Kimbrough, 552 U. S., at 103, one does not go into a conversation with a dietitian under the assumption that all foods not expressly prohibited are on the table. Those concerned enough about their diets to seek professional help likely do not need to be expressly told that they should avoid “[f]ried chicken.” Ante, at 12.

 Congress’s instructions to the Commission are much the same. An instruction not to grant compassionate release based on rehabilitation alone is not a prohibition against considering any other type of evidence. Instead, courts may consider all relevant information, as they have always done, see Concepcion, 597 U. S., at 491–492, and the Commission may authorize them to do so, as it did here.

 First, the majority argues that considering nonretroactive changes in law was, “[u]ntil very recently,” unprecedented. Ante, at 10. Far from it. Again, courts have for centuries considered all relevant evidence at sentence-modification proceedings, and sentencing disparities like those created by nonretroactive changes in law are relevant information. See Concepcion, 597 U. S., at 490–492; supra, at 10–12.

 The majority offers a counter-tradition, claiming that the “the heartland of compassionate release” “has long been defined by a prisoner’s personal circumstances,” such as his physical health. Ante, at 10. It is certainly true that personal circumstances feature prominently in compassionate-release proceedings. Even so, nothing in the statute’s text confines courts to considering only personal circumstances, and “[t]he only limitations on a court’s discretion to consider any relevant materials” in a sentence-modification proceeding “are those set forth by Congress in a statute or by the Constitution.” Concepcion, 597 U. S., at 494. Without an express limitation, the default remains: Courts may consider facts beyond personal circumstances when deciding whether to modify a defendant’s sentence.

 Nor does historical practice supply a personal-circumstances limit. The parties have identified only two published opinions in which district courts reduced a sentence under the predecessor to the compassionate-release statute, which was in effect from 1976 to 1984. In one of those two decisions, the District Court reduced a defendant’s sentence not based on the personal circumstances of the  defendant, but rather because of the disparity between the defendant’s sentence and those of his codefendants. See United States v. Diaco, 457 F. Supp. 371, 372, 376 (D NJ 1978); see also United States v. Banks, 428 F. Supp. 1088, 1089–1090 (ED Mich. 1977) (granting relief under predecessor statute “upon careful review of the prisoner’s record,” including evidence of rehabilitation).

 Further, this Court has also acknowledged that, apart from a defendant’s personal circumstances, compassionate release might be appropriate to reduce a sentence to account for later legal developments. In fact, in Setser v. United States, 566 U. S. 231 (2012), the Court addressed whether federal courts have authority to order federal sentences to run consecutively to anticipated state sentences that have not yet been imposed. Id., at 233. After holding such sentences to be permissible, Justice Scalia explained that, when unexpected sentencing developments in state court “produc[e] unfairness to the defendant,” the Sentencing Reform Act “provides a mechanism for relief ”: compassionate release under §3582(c)(1)(A). Id., at 242–243. These examples confirm that courts have historically considered postsentencing legal developments beyond a defendant’s personal circumstances.²

 Without a foothold in the compassionate-release statute or its history, the majority looks for one in the First Step Act instead. According to the majority, Congress’s decision not to make its changes to §924(c) retroactive means that it intended to preclude consideration of those changes in all cases as part of an individualized compassionate-release  determination. That argument badly overreads the First Step Act.

 To start, Congress’s decision not to make its changes retroactive simply means that Congress did not intend for every person who committed §924(c) offenses prior to the First Step Act’s passage to be automatically eligible for relief. See ante, at 14 (“True, Congress chose not to reduce §924(c) penalties on a categorical basis”). By making that one choice, however, Congress did not also decide that sentencing disparities created by those changes should never be considered in deciding compassionate-release motions. Indeed, Congress amended the compassionate-release statute in the First Step Act as well, so Congress easily could have specified that its nonretroactive changes should not be considered in compassionate-release determinations, just as it previously limited consideration of rehabilitation. See Concepcion, 597 U. S., at 494 (“Congress is not shy about” limiting courts’ discretion). Yet Congress did not do so.

 The majority would have a point if the Commission instructed courts to grant relief based solely on a nonretroactive change in law; such an approach would amount to granting relief on a categorical basis, which would be in tension with the categorical nonretroactive nature of Congress’s amendments here. The Commission’s guidance, however, permits consideration of changes in law only when the disparity is “gross,” the sentence is “unusually long,” and the defendant has served at least 10 years of his sentence. §1B1.3(b)(6). It also requires courts to give “full consideration of the defendant’s individualized circumstances” while weighing any change in law. Ibid. That individualized approach is consistent with Congress’s judgment that prior §924(c) offenders are not automatically entitled to relief, and in no way countermands Congress’s more general decision not to make the First Step Act’s amendments retroactive for each and every §924(c) offender.

  Take petitioners as examples. Neither is seeking relief based solely on Congress’s nonretroactive changes to §924(c). Instead, they argue that they are entitled to a reduction because of the change in law in addition to their remarkable changed personal circumstances. To be clear, it may well be that neither Rutherford nor Carter would be found by a district court to be entitled to compassionate release even after taking account of sentencing disparities. Nevertheless, they are at least entitled to full consideration of their cases in view of all relevant information. Affording them that individualized analysis is fully consistent with Congress’s judgment that a change in law does not automatically entitle them to relief.

 Even the Government agrees that individualized analysis is consistent with Congress’s intent at least at one step of the compassionate-release analysis. Once a defendant establishes extraordinary and compelling reasons, the Government concedes that courts may consider a sentencing disparity created by nonretroactive changes in law when weighing the §3553(a) factors to decide how much to reduce a defendant’s sentence. See Brief for United States 37; see also 18 U. S. C. §3553(a)(6) (listing “the need to avoid unwarranted sentence disparities” as one factor the court “shall consider”). The majority maintains that considering sentencing disparities is impermissible at the first step because it is a “distinct analytical step that imposes independent and ascertainable limits on access to compassionate release.” Ante, at 14.

 This attempt at statutory surgery does not cure the problem. For one thing, §3582(c)(1)(A)(i) does not clearly mandate that the two steps be hermetically sealed off from one another. It specifies only that a court “may reduce [a] term of imprisonment . . . , after considering the factors set forth in section 3553(a) to the extent that they are applicable, if it finds that . . . extraordinary and compelling reasons warrant such a reduction.” §3582(c)(1)(A)(i). This text in no  way dictates the majority’s precise, two-step sequencing in which courts undertake two different inquiries while feigning ignorance of the other. That text just as easily means that courts should decide if there are extraordinary and compelling reasons for relief “after considering” the §3553(a) factors. Indeed, it is obvious, and conceded, that at least some of the §3553(a) factors are salient at both stages. For instance, one of the §3553(a) factors is “the history and characteristics of the defendant,” §3553(a)(1), which includes the defendant’s personal circumstances. If courts can consider personal circumstances at both steps of the analysis, see ante, at 10 (describing “personal circumstances” as “the heartland of compassionate release”), it is not clear why they cannot also consider sentencing disparities at both steps as well.

 In any event, even assuming Congress intended to impose a rigid, two-step process, it is hard to “see how a court may be thought to subvert congressional intent by considering nonretroactive changes to the law at the ‘extraordinary and compelling’ stage of the analysis but not while weighing the §3553(a) factors.” United States v. Ruvalcaba, 26 F. 4th 14, 32 (CA1 2022) (Barron, J., concurring). “[I]f Congress truly intended to bar district courts from considering [the First Step Act’s] changes to mandatory minimums in the compassionate release context by making the changes non-retroactive, then it is doubtful those changes should be considered at all.” United States v. Chen, 48 F. 4th 1092, 1099 (CA9 2022). It also remains that, no matter how the Court may try to slice it, a compassionate-release proceeding is a “sentence-modification proceeding,” and courts have long enjoyed broad discretion to consider relevant information in such proceedings unless Congress expressly says otherwise. Concepcion, 597 U. S., at 491. Nothing in the First Step Act specifically addresses what courts may consider when deciding whether a defendant is eligible for a sentence reduction.

  In the end, the majority is unable to explain why exactly the Commission’s case-by-case approach is inconsistent with Congress’s categorical nonretroactivity decision. See ante, at 14–15. After acknowledging that it is “[t]rue” that Congress merely made a judgment about categorical retroactivity, the majority takes aim at something different altogether: mandatory minimums. See ibid. In the majority’s view, permitting holistic consideration of all relevant evidence, including changes in law, would also permit judges to grant relief based on their personal view that a mandatory-minimum sentence is too harsh. Ante, at 15.

 Those fears are unfounded. A judge’s policy disagreement with a statutory mandatory minimum is not, and has never been, a relevant consideration at sentencing. Although district courts may disagree with the nonbinding Sentencing Guidelines “ ‘based solely on policy considerations,’ ” they are “constrained by the mandatory minimums Congress prescribed” by statute regardless of their policy views. Kimbrough, 552 U. S., at 101, 108; see United States v. Fanfan, 558 F. 3d 105, 110, n. 6 (CA1 2009) (“Though district courts may sentence based on policy disagreements with the . . . guidelines, they may not do so on the basis of disagreements with statutes”); see also §3553(e) (providing “[l]imited authority” for a court to “impose a sentence below a statutory minimum”); §3553(f ) (permitting district courts to disregard a mandatory minimum for certain offenses only if five criteria are met).

 Reflecting that norm, §3553(a) does not permit judges to consider their own personal policy preferences when imposing an initial sentence or reducing a sentence through a compassionate-release motion. Nor does it allow judges to weigh whether they think that Congress mandated the right punishment for a given crime. It does, however, instruct judges to consider sentencing disparities, §3553(a)(6), including those created by nonretroactive changes in law. That instruction is perfectly consistent  with courts’ longstanding discretion to consider all relevant information, including changes in law, when resentencing a defendant. See Concepcion, 597 U. S., at 491–492.

 In addition, no evidence supports the specter that the majority invokes. District judges around the country routinely impose mandatory-minimum sentences in tens of thousands of cases every year, as it is their oath-bound duty to apply the laws of the United States. See 28 U. S. C. §453. There is no evidence that these same district judges are routinely granting compassionate release as an end run around mandatory minimums. In Fiscal Year 2024, the Commission received information regarding 61,678 cases in which a criminal defendant was sentenced. United States Sentencing Commission, QuickFacts on Mandatory Minimums 1 (2024), https://www.ussc.gov/sites/default/files/pdf/ research-and-publications/quickfacts/Quick_Facts_Mand_ Mins_FY24.pdf (archived at https://perma.cc/5Q4L-9G3Z). Of all individuals sentenced that year, 15.9% faced a mandatory minimum at sentencing. That same year, judges granted compassionate release just 481 times in total on any ground (including under §1B1.13(b)(6)) and denied it 2,534 times. Compassionate Release Report, Table 20. Compassionate release is not the loophole to mandatory minimums that the majority fears.

 Even if a district court were to start using compassionate release as an opportunity to usurp Congress’s role and elevate its own policy preferences, the Government could appeal those outlier grants of compassionate release to the courts of appeals and, if necessary, to this Court. If a systemic solution were necessary, the Commission could issue another policy statement or Congress could enact legislation to prevent misuse of the compassionate-release statute. The solution is not, however, for this Court to conjure categorical limitations on courts’ discretion that neither Congress nor the Commission imposed.

 The majority is right on one point: The extraordinary-and-compelling analysis is indeed subject to “independent and ascertainable limits.” Ante, at 14. Yet one need only read the Commission’s policy statement to determine those limits. The majority nevertheless searches high and low for other statutory limits on what courts may consider when deciding compassionate-release motions. Because the Commission’s narrow, limited guidance is consistent with what Congress has previously said and with what courts have previously done, I respectfully dissent.