After Esteras v. United States: Is It Time to Amend 18 U.S.C. § 3583

With substantial early release now again part of federal corrections law and practice, indeterminacy is back, and the original offense and sentence are newly relevant at supervised release revocations

This is the second in a series of essays about sentencing and corrections law cases decided by the Supreme Court during its October 2024 term.

- - -

The Sentencing Reform Act of 1984 (SRA) dramatically changed federal sentencing and corrections law. With its enactment, Congress rejected the system of indeterminate sentencing that had been in place for much of the country’s history. Under indeterminate sentencing, when a federal judge imposed an imprisonment term as part of an offender’s sentence, say five years, it was unclear at the time of sentencing — not determined — how much of that five year imprisonment term the offender would actually serve. The offender’s prison release date would be fixed long after sentencing by the United States Parole Commission, a Justice Department agency, based on its assessment of when the person had been rehabilitated and was ready for release. Before the SRA, most offenders sentenced to federal prison were eligible for parole once they had served one-third of the imposed imprisonment sentence, and most would be released after serving about half.

The SRA abolished that system. One of the law’s goals was to bring “truth in sentencing” to the federal criminal justice system. The Act eliminated parole release and created the U.S. Sentencing Commission to develop sentencing guidelines that would establish a range of determinate sentences for various categories of offenses and offenders. What made the sentences “determinate” was that offenders would serve the imposed sentence in its entirety, less at most 15% for good behavior in prison. (Stephen Colbert might call this, more accurately, “truthiness-in-sentencing.”)

Along with these changes in sentencing law, the SRA also established supervised release. Unlike parole release, which was based on the idea of offenders serving a significant part of their original sentence in the community, supervised release was not intended to be part of the original sentence. The Sentencing Commission characterizes it as “a separate order of supervision in addition to any term of imprisonment imposed by the court,” U.S. Sent’g Comm, Guidelines Manual, Chapter 7, Part A. It is a form of post-imprisonment supervision which the Supreme Court recognized in United States v. Johnson, 529 U.S. 53, 59 (2000), serves only as a means of rehabilitation, distinct from the goals of incarceration.

The legislative history of the SRA also makes clear that supervised release is separate and distinct from the original sentence. It was designed “to ease the defendant’s transition into the community after the service of a long prison term for a particularly serious offense, or to provide rehabilitation to a defendant who has spent a fairly short period in prison for punishment or other purposes but still needs supervision and training programs after release.” S. Rep. No. 98-225 at 124 (1983). Supervised release comes with conditions of behavior to further its transition and rehabilitative goals. And to give the conditions teeth, if an offender violates the conditions of release, that release can be revoked and additional sanctions imposed, including an additional term of imprisonment.

- - -

With the development of the new determinate sentencing system and the distinct goals of supervised release, it made perfect sense for Congress to limit the factors a court could consider in determining whether to include a term of supervised release, the length and conditions of such a term, and whether to revoke such a term. This limitation is contained in 18 U.S.C. § 3583, and it was affirmed by the Supreme Court in Esteras v. United States, 606 U.S. ___ (2025), decided a week or so ago. In a 7-2 decision authored by Justice Barrett, the Court ruled that because supervised release “fulfills rehabilitative ends” and is intended only to provide “individuals with postconfinement assistance,” when determining the sanction for a supervised release violation, a court must only consider “forward-looking ends of sentencing (deterrence, incapacitation, and rehabilitation), but may not consider the backward-looking purpose of retribution.” (Emphasis in original.)

When a defendant is first sentenced for a federal crime, courts are instructed to consider the so-called § 3553(a) factors, which include, in § 3553(a)(2)(A), the need for the sentence to reflect the seriousness of the offense, promote respect for the law, and provide just punishment. Justice Barrett, in her majority opinion in Esteras, summed up these factors as “retribution,” a backward-looking purpose of sentencing. And 18 U.S.C. § 3583, the statute listing the factors sentencing courts can consider for supervised release imposition and violations, omits these specific retributive factors from the revocation consideration.

In Esteras, the petitioners – Edgardo Esteras, Timothy Jaimez, and Toriano Leaks – were each sent back to prison for breaking supervised release rules, based in part on district court rulings that cited punishment, among other factors, as justifying a prison term for the rules’ infractions. The Sixth Circuit said that was ok, despite the omission of the retributive factors – punishment – from 18 U.S.C. § 3583. The Supreme Court reversed the Sixth Circuit, holding that congressional intent and the principles of statutory construction require that in sentencing someone for a revocation, courts not consider the seriousness of the original offense, or promoting respect for the law, or providing just punishment related to the original offense – the retributive factors.

Congress’s decision to enumerate most of the sentencing factors while omitting
§ 3553(a)(2)(A) raises a strong inference that courts may not consider that factor when deciding whether to revoke a term of supervised release. The inference is consistent with both the statutory structure and the role that supervised release plays in the sentencing process.

There can be no doubt that the inference is indeed consistent with the role that supervised release played in the sentencing process when the SRA was passed. The basic principles underlying the 1984 sentencing reform was first, to create a tariff-based sentencing system – if you commit a particular crime, you generally are sentenced in accord with the severity of that crime (the tariff) – and second, to require offenders to serve the sentence imposed, less a limited amount of good time. Supervised release came after the offender served the nearly complete term of imprisonment. In such a sentencing system, it would be improper to reconsider retribution for the original offense for any supervised release revocation. That retribution had already been meted out at the original sentencing and had been served prior to release.

But what the Court does not grapple with in Esteras is that in the years since the SRA, and especially since the passage of the First Step Act in 2018, federal sentencing has been slowly moving away from the determinate system envisioned by Congress in 1984. This is no slight on Justice Barrett and her opinion; as a textualist, her job as she saw it was simply to interpret § 3583 as written. But in considering the policy embodied in the statute, it is important to recognize that under current law, a substantial percentage of offenders do not serve the imprisonment sentence imposed less good conduct time. Now, with good time credits, First Step Act earned time credits for completing recidivism reducing programming or productive activity, perhaps a sentence reduction for completing residential drug treatment, and maybe even a sentence reduction for newly expanded compassionate release reasons too, a defendant may be released after having served less than half of the imposed imprisonment sentence. In this system, the retribution that had been meted out at the original sentencing has not always been served.

When an offender is released before completing the sentence that reflects the seriousness of the offense and provides just punishment, it might, in such circumstances, be quite appropriate to consider the retributive goals if she violates the terms of supervised release. Retribution might have already been meted out at the original sentencing, but in the cases of substantially early release, it would not have been served. The more indeterminate federal sentencing system we have now requires a reexamination of the factors for supervised release consideration listed in § 3583.

For example, suppose Elizabeth Holmes, the biotech entrepreneur who was convicted of felony fraud charges stemming from her blood-testing company, Theranos, is released from prison after serving seven years of the approximately 11-year sentence she received after her conviction. And suppose she begins developing a new fraud scheme but is turned in before anything comes of it. Would it be sound sentencing policy to allow a judge, upon revocation of Holmes’ supervised release, to consider reimposing a part of the earlier-imposed prison sentence that went unserved as a sanction for the supervised release term and, in so doing, to vindicate the retributive goals embodied in the original sentence? For decades, parole revocation often resulted in requiring the offender to serve the remaining portion of the originally imposed imprisonment term. If nothing else, it vindicated the retributive interests. In addition to furthering the crime control and rehabilitative purposes — the forward-looking ends of sentencing — it seems appropriate to reconsider the retributive purposes — the backward-looking ends of sentencing — when supervised release is revoked and when those purposes have not yet been met.

Of course, all of this fuss over purposes and factors are what Justice Alito, in his dissent in Esteras, calls the “world of airy abstractions.” It probably has quite little practical effect one way or the other. Justice Barrett’s majority opinion actually provides a roadmap of sorts to sentencing judges to consider the original offense as part of the revocation process and to avoid running afoul of the limits of § 3583. But as Justice Barrett, herself, states at the end of the opinion, this does not mean that the decision “amounts to a substance-free reverse magic-words requirement.” The principle of limiting the scope of retribution at supervised release revocation is real, codified, and should be taken seriously. It is “very much substantive.” Under current law, Justice Barrett writes, “[d]istrict courts may not consider the retributive purpose of § 3553(a)(2)(A) before revoking supervised release. We trust that district courts will heed that instruction regardless of the practical likelihood of reversal.”

But maybe too, Congress ought to rethink 18 U.S.C. § 3583 and its limitations on the factors to be considered upon a supervised release revocation. Maybe Congress should allow district courts to consider retributive goals when an offender violates the terms of supervised release and has not fully served the previously imposed imprisonment term. With substantial early release now routinely part of federal sentencing and corrections law and practice, indeterminacy is back. And with it, the original offense and sentence seem newly relevant at supervised release revocations.