Ellingburg Exposes Federal Restitution's Central Contradiction

The Supreme Court now calls MVRA restitution criminal punishment. Yet the federal system still imposes it without regard to ability to pay and generally denies defendants an ordinary collateral remedy to challenge it.

Restitution begins with an appealing principle: a person who causes financial harm should repay the victim. But federal restitution has evolved into something more complicated. It is presented as compensation, imposed as part of a criminal sentence, enforced through the power of the federal government, and often carried for decades by defendants who have little realistic prospect of paying the full amount.

Victims may be promised compensation that never arrives. Defendants may leave prison only to remain subject to liens, garnishment, financial disclosure, collection proceedings, and payment obligations that continue long after incarceration and supervised release. In many cases, restitution operates less like a realistic repayment plan and more like a permanent financial sentence.

The Supreme Court has now acknowledged the punitive reality of that system.

In Ellingburg v. United States, 607 U.S. 163 (2026), a unanimous Court held that restitution imposed under the Mandatory Victims Restitution Act of 1996, or MVRA, is "plainly criminal punishment" for purposes of the Ex Post Facto Clause.

That holding is significant—but not limitless.

Ellingburg establishes that the government cannot avoid constitutional scrutiny by characterizing MVRA restitution as merely compensatory. At the same time, it does not automatically permit defendants to challenge restitution through 28 U.S.C. § 2255. It does not make every restitution statute punitive, eliminate the finality of restitution judgments, authorize the government to waive mandatory restitution, or decide that every criminal-punishment doctrine now applies to restitution.

The result is a striking legal contradiction:

Federal restitution is criminal punishment when the Constitution restricts its retroactive imposition, but it ordinarily remains noncustodial when a defendant seeks collateral review under § 2255.

Understanding Ellingburg therefore requires examining both sides of the decision: what the Court changed and what the existing law still prevents defendants from doing.

The Promise and Reality of Federal Restitution

Congress enacted the MVRA to require restitution for designated federal offenses. For covered crimes, the sentencing court generally must order restitution in the full amount of each victim's legally compensable losses.

The statute draws an important distinction between the amount owed and the method of payment.

When calculating the amount, the court must order the full qualifying loss "without consideration of the economic circumstances of the defendant." Only after determining that amount does the court consider the defendant's assets, projected income, obligations, and dependents when establishing the payment schedule. A court may therefore recognize that a defendant cannot realistically pay the judgment while still being required to enter the full amount on paper.

That structure helps explain why restitution orders can become enormous and largely uncollectible.

A 2025 report from the National Association of Criminal Defense Lawyers—an advocacy publication drawing substantially from government data—cited a Government Accountability Office finding that approximately $110 billion in federal restitution debt remained outstanding, with about $100 billion classified as uncollectible. The report also stated that the government had collected the full restitution amount in only about five percent of cases covered by the underlying data.

The same report placed the average federal restitution order at more than $3.3 million and found that federal agencies were identified as victims in almost 40 percent of the cases reflected in the relevant data. Those figures complicate the conventional image of restitution as a straightforward transfer from an offender to an individual crime victim.

Federal law also supplies powerful enforcement mechanisms. Restitution may be enforced through liens, garnishment, property collection, and other civil collection remedies. Liability generally continues until the later of twenty years after entry of judgment or twenty years after the defendant's release from imprisonment. A material change in financial circumstances may justify modifying the payment schedule under § 3664(k), but that provision ordinarily does not authorize the court to erase or recalculate the underlying restitution amount.

A defendant cannot be imprisoned solely because genuine indigence makes payment impossible. But knowing or willful default can trigger modification or revocation of supervision, contempt proceedings, adjustment of the payment schedule, and, under limited statutory conditions, resentencing or imprisonment.

These are not merely the consequences of a private debt. They are exercises of governmental authority arising from a criminal judgment.

That reality became central in Ellingburg.

What the Supreme Court Decided in Ellingburg

Holsey Ellingburg committed his offense before the MVRA became law on April 24, 1996. He was sentenced later in 1996 and ordered to pay $7,567.25 in restitution under the new statutory regime.

Years later, Ellingburg challenged the government's continued enforcement of the obligation. He argued that applying the MVRA to conduct predating its enactment violated the Ex Post Facto Clause.

The Eighth Circuit rejected his argument because circuit precedent treated MVRA restitution as nonpunitive. The Supreme Court unanimously reversed.

The Court treated the classification question as one of statutory construction. It examined the MVRA's text, structure, placement, procedures, and relationship to the criminal sentencing process.

Several statutory features drove the Court's conclusion:

• The MVRA calls restitution a "penalty" for a criminal "offense."
• It applies only after the conviction of a criminal defendant.
• Restitution is imposed during sentencing alongside imprisonment and fines.
• For certain misdemeanors, restitution may be imposed in place of other criminal penalties.
• The United States—not the individual victim—is the party adverse to the defendant.
• The restitution provisions appear in Title 18 and in chapters governing criminal sentences.
• Courts use criminal sentencing procedures when imposing restitution.
• Knowing nonpayment may lead to modification of supervision or additional sanctions.

Taken together, these features made the statutory analysis straightforward: MVRA restitution is criminal punishment.

The Court did not ignore restitution's compensatory function. Instead, it concluded that the MVRA has dual purposes. Congress intended restitution both to compensate victims and to punish defendants. Once the statute's text and structure demonstrated a punitive purpose, the existence of a compensatory purpose did not convert the sanction into a civil remedy.

The Court also emphasized that victims do not control the restitution proceeding as they would control a private civil lawsuit. A victim does not initiate the criminal prosecution, cannot privately settle the government's restitution demand, and is not the party prosecuting the defendant. Those features reinforce restitution's public and punitive character.

Ellingburg Does Not Mean That Every Restitution Law Is Criminal

The Court expressly limited its holding to the text and structure of the MVRA.

It did not announce that every statute bearing the label "restitution" necessarily imposes criminal punishment. Another statutory scheme could theoretically operate as a civil remedy if its text, enforcement structure, parties, and procedures were materially different.

The Court also cautioned that not everything appearing in a criminal judgment—or everything occurring at sentencing—is automatically punishment. The holding applies to MVRA restitution because of the combined features of that particular statutory regime.

That limitation matters. Ellingburg supplies an analytical method, not a universal label. Courts examining other sanctions must still evaluate the governing statute as a whole.

What Ellingburg Means for Ex Post Facto Challenges

The Ex Post Facto Clause prevents the government from retroactively imposing or increasing criminal punishment for conduct completed before the governing law took effect.

Before Ellingburg, defendants challenging the retroactive use of the MVRA faced a threshold obstacle: some courts treated restitution as civil or compensatory and therefore outside the Ex Post Facto Clause.

The Supreme Court removed that obstacle.

For conduct occurring before April 24, 1996, a defendant may now rely on Ellingburg to argue that the government cannot retroactively apply materially harsher MVRA provisions without satisfying the Constitution's prohibition against ex post facto punishment.

That does not mean every pre-MVRA restitution order is automatically invalid. A defendant must still establish, among other things, that:

• The challenged obligation was actually imposed or extended under the MVRA;
• The relevant conduct occurred before the applicable provision took effect;
• The later law created a sufficiently disadvantageous or more burdensome punishment;
• The claim is presented through a procedurally available vehicle; and
• No independent waiver, forfeiture, finality, or other procedural defense defeats the claim.

The Supreme Court itself did not direct entry of judgment eliminating Ellingburg's obligation. It reversed and remanded, permitting the lower court to consider the government's separate arguments for affirmance.

Thus, Ellingburg resolves the punishment question. It does not resolve every remaining issue in an individual case.

The Broader Constitutional Significance

The majority opinion was narrow in doctrinal terms, but its reasoning may influence a broader debate over financial sanctions.

Restitution is no longer accurately described as civil compensation that merely happens to appear in a criminal judgment. The Supreme Court has now recognized it as part of the punishment imposed by the sovereign for a public wrong.

That recognition may strengthen arguments that restitution proceedings deserve protections commensurate with their punitive consequences. Those arguments may involve due process, the defendant's presence at critical sentencing proceedings, the reliability of loss calculations, proportionality, notice, the Excessive Fines Clause, or the procedures governing post-sentencing restitution determinations.

But those questions remain separate. Ellingburg did not hold that every restitution order is an excessive fine. It did not constitutionalize an ability-to-pay requirement. It did not determine that every restitution hearing is a critical stage requiring the defendant's presence. And it did not establish a new standard for evaluating the proportionality of restitution.

The decision makes those arguments more difficult to dismiss as attacks on a merely civil remedy. It does not decide them.

Justice Thomas's Concurrence Points Beyond Restitution

Justice Thomas, joined by Justice Gorsuch, wrote separately to question the Court's modern framework for distinguishing civil regulation from criminal punishment.

The concurrence argued that the original understanding of the Ex Post Facto Clauses, reflected in Calder v. Bull, focused on whether the sovereign imposed a coercive penalty to redress a public wrong—not on whether the legislature formally labeled the proceeding "civil" or placed the law in a civil code.

Under that approach, a financial sanction could constitute punishment even when enforced through nominally civil procedures. The critical questions would be what the government is doing, whose injury it is vindicating, and whether it is imposing a coercive deprivation of life, liberty, or property for a public offense.

That concurrence is not controlling law. But it suggests that Ellingburg may become relevant beyond conventional criminal restitution. Future cases could invoke its reasoning when analyzing civil penalties, administrative monetary sanctions, disgorgement, forfeiture, or other government-imposed financial consequences.

The majority did not go that far. Two Justices expressly signaled interest in doing so.

The Central Paradox: Punishment Does Not Automatically Mean "Custody"

The recognition of restitution as criminal punishment naturally raises another question:

If restitution is punishment imposed as part of a criminal sentence, why can a federal prisoner not ordinarily challenge it under 28 U.S.C. § 2255?

The answer lies in the text of § 2255. The statute authorizes relief for:

"A prisoner in custody under sentence of a court established by Act of Congress claiming the right to be released."

Section 2255 is not a general mechanism for correcting every unlawful feature of a federal criminal judgment. It is a collateral remedy centered on custody and release.

A prison sentence satisfies that requirement. Parole and supervised release may also constitute custody because they significantly restrain liberty.

An ordinary restitution order, however, imposes a monetary obligation. Vacating or reducing it generally does not release the prisoner from incarceration, supervision, or another qualifying restraint.

That is the distinction Ellingburg did not address:

A sanction may be criminal punishment without being custodial punishment.

The Predominant Rule Across the Federal Circuits

The exclusion of ordinary restitution claims from § 2255 is not the rule of only one federal circuit.

Courts of appeals across the country have repeatedly concluded that § 2255 does not ordinarily authorize relief directed solely at restitution. Representative decisions include:

Smullen v. United States, 94 F.3d 20, 25–26 (1st Cir. 1996); Kaminski v. United States, 339 F.3d 84, 87–91 (2d Cir. 2003); United States v. Hatten, 167 F.3d 884, 887 (5th Cir. 1999); Barnickel v. United States, 113 F.3d 704, 706 (7th Cir. 1997); United States v. Bernard, 351 F.3d 360, 361 (8th Cir. 2003); United States v. Kramer, 195 F.3d 1129, 1130 (9th Cir. 1999); and Blaik v. United States, 161 F.3d 1341, 1342–43 (11th Cir. 1998).

The courts do not always describe the limitation in precisely the same way.

Some characterize the custody requirement as jurisdictional and state that the district court lacks subject-matter jurisdiction over a restitution-only claim. Others treat the issue as one of statutory scope or cognizability. The practical outcome is usually the same: § 2255 does not ordinarily authorize an order vacating or reducing restitution when the requested relief would not affect custody.

The distinction in terminology matters. It is more accurate to describe the rule as a predominant national rule of noncognizability than to assert that every circuit uses an identical subject-matter-jurisdiction analysis.

A Valid Prison Claim Does Not Automatically Carry the Restitution Claim

A defendant ordinarily cannot overcome the limitation by combining a restitution challenge with a valid challenge to imprisonment.

Courts generally examine each claim according to the relief it seeks. One count of a § 2255 motion may properly challenge a prison term, while another count directed only at restitution remains outside the statute.

In other words, a noncustodial claim does not become custodial through procedural packaging.

The Eleventh Circuit has recently reiterated that habeas relief does not extend to noncustodial portions of a sentence merely because they appear in the same petition as custodial claims. The Second and Ninth Circuits have applied substantially the same reasoning.

A successful challenge to the underlying conviction may sometimes affect restitution indirectly. If the conviction supporting the restitution order is vacated and the defendant receives a new judgment or resentencing, multiple parts of the sentence may change. That is different from using § 2255 to obtain relief on a claim that targets only the amount or existence of restitution.

A Timely Motion May Still Be Noncognizable

Timeliness and statutory cognizability are separate questions.

Section 2255(f) contains a one-year limitations period. Under § 2255(f)(3), the period may begin when the Supreme Court initially recognizes a new right that has been made retroactively applicable to cases on collateral review.

A defendant may therefore argue that Ellingburg created a newly recognized constitutional right relevant to the defendant's restitution obligation.

But even assuming the motion is timely, the defendant must still satisfy § 2255(a). A filing deadline determines when an authorized claim may be brought. It does not expand what claims the statute authorizes.

Ellingburg did not decide:

• Whether its holding applies retroactively through § 2255(f)(3);
• Whether a restitution challenge constitutes a claim for release from custody;
• Whether restitution itself satisfies the custody requirement; or
• Whether § 2255 reaches every criminal punishment contained in a judgment.

Accordingly, a timely § 2255 motion directed only at restitution ordinarily encounters the same custody barrier as an untimely one.

The Sixth Circuit's Narrow Qualification

The circuit rule is broad, but it is not perfectly uniform in every doctrinal detail.

In Weinberger v. United States, 268 F.3d 346, 351 n.1 (6th Cir. 2001), the Sixth Circuit permitted a particular ineffective-assistance-of-counsel claim involving restitution. The court did not broadly hold that restitution constitutes custody.

The Sixth Circuit later characterized Weinberger as a special rule involving ineffective assistance and reaffirmed that a petitioner seeking only relief from a monetary assessment ordinarily is not claiming a right to be released under § 2255. Amaya v. United States, No. 21-1781 (6th Cir. June 23, 2023).

The appropriate conclusion is therefore not that § 2255 can never have any connection to restitution under any circumstances. It is that the overwhelming weight of appellate authority excludes ordinary restitution-only claims, subject to limited circuit-specific qualifications.

Miclaus Confirms the Limits of Ellingburg

The Sixth Circuit's 2026 decision in United States v. Miclaus illustrates why Ellingburg must be applied to the issue it actually decided.

Miclaus arose on direct appeal following resentencing, not through a § 2255 motion. The defendant argued that because Ellingburg identified the government as the adverse party during restitution proceedings, the government should have the authority to waive mandatory restitution.

The Sixth Circuit rejected that inference.

The court explained that Ellingburg did not address waiver or the permissible scope of resentencing following a general remand. The fact that the government is the adverse party for purposes of the punishment analysis does not permit it to waive restitution that Congress made mandatory. Nor does it narrow the district court's resentencing authority after a general remand.

Miclaus did not decide whether restitution is cognizable under § 2255. Its broader lesson is that the word "punishment" does not automatically rewrite every separate statute or procedural doctrine touching restitution.

Section 2255 Is Not the Only Possible Procedural Vehicle

The unavailability of ordinary § 2255 relief does not necessarily mean that every restitution challenge is impossible. The proper vehicle depends on the nature of the alleged error, the procedural history, the requested remedy, the governing circuit, and whether the judgment is final.

A challenge to the original legality or amount of restitution ordinarily should be raised at sentencing and on direct appeal.

Section 3664(k) may permit adjustment of the payment schedule following a material change in the defendant's economic circumstances. It generally addresses how restitution is paid, not whether the original amount was legally valid.

The finality provisions of § 3664(o) identify other limited mechanisms through which a restitution judgment may be corrected, appealed, amended, adjusted, or affected by resentencing. Those provisions are specific and do not give district courts unrestricted authority to reopen restitution judgments whenever equitable concerns arise.

Extraordinary common-law writs may also be argued in unusual cases. In Ramdeo v. United States, 136 F.4th 1348 (11th Cir. 2025), the Eleventh Circuit held that custody does not categorically prevent a petitioner from seeking coram nobis relief against a noncustodial component of a sentence. The court did not hold that coram nobis necessarily could or should vacate the restitution order; it remanded while emphasizing the writ's demanding requirements.

That emerging authority supports a careful formulation:

Section 2255 ordinarily does not provide a remedy for restitution, but the absence of a § 2255 remedy is not always equivalent to the absence of every conceivable remedy.

Any alternative remains highly dependent on procedural posture and circuit precedent.

Ellingburg Reveals a Deeper Policy Failure

The most important consequence of Ellingburg may extend beyond litigation over offenses committed before 1996.

The decision forces the federal system to speak honestly about what restitution is.

For years, restitution has been justified primarily through the language of victim compensation. But the MVRA requires full-loss judgments even when the defendant has no realistic capacity to satisfy them. The result may be an enormous judgment that remains unpaid, disappoints victims, and subjects the defendant to decades of collection and financial oversight.

The NACDL report argues that this system often leaves both sides dissatisfied. Victims receive formal judgments rather than actual repayment, while defendants remain financially entangled with the criminal system after completing the custodial portions of their sentences.

That is not an argument against accountability. It is an argument for distinguishing meaningful accountability from symbolic debt.

A smaller, realistically collectible order may provide a victim with more actual compensation than a multimillion-dollar judgment that will never be paid. A transparent payment structure may generate greater compliance than an obligation the defendant understands to be impossible from the day it is imposed.

If restitution is punishment, then the amount, procedure, and enforcement of that punishment should receive the same seriousness as other parts of sentencing.

What Meaningful Reform Could Include

The Court did not order Congress to reform the MVRA. But Ellingburg strengthens the policy case for reconsidering several features of federal restitution.

Ability to Pay Should Matter to More Than the Payment Schedule

Current law generally separates the full amount of restitution from the defendant's financial circumstances. Congress could permit courts to consider realistic earning capacity when determining what restitution can actually accomplish.

That would not require abandoning victim compensation. It would require acknowledging that a paper judgment beyond any conceivable ability to pay may compensate no one.

Restitution Should Be Treated as a Critical Part of Sentencing

Restitution calculations can involve causation, loss methodology, future expenses, offsets, insurance, interest, and the identification of legally qualifying victims. Those questions should not be treated as administrative details after imprisonment has already been decided.

Defendants should receive meaningful notice, access to the supporting loss evidence, an opportunity to be heard, and clear explanations of the long-term enforcement consequences.

Orders Should Reflect Actual, Legally Attributable Loss

Reform could more closely connect restitution to losses proximately caused by the offense of conviction. Congress could also address concerns involving joint-and-several liability, overlapping recovery, interest, penalties, and compensation for losses that may already have been reimbursed by other sources.

The System Should Measure Actual Victim Recovery

A restitution system cannot be evaluated solely by the amount ordered. It must also measure how much reaches victims, how long payment takes, which categories of recipients receive funds, and how much public money is spent attempting to collect debts already classified as uncollectible.

Without reliable data, the system risks confusing the appearance of accountability with its actual results.

Ellingburg Is Both Broader and Narrower Than It First Appears

The decision is broader than a technical dispute involving a $7,567.25 restitution order from 1996. It establishes that the MVRA's combination of punitive language, criminal procedures, government enforcement, and sentencing consequences brings restitution within the constitutional category of criminal punishment.

But the decision is narrower than a general declaration that every restitution order is now open to collateral attack.

Ellingburg does not itself:

• Make restitution "custody" under § 2255;
• Authorize a restitution claim merely because the defendant remains incarcerated;
• Permit a noncustodial claim to accompany a custodial one through procedural bootstrapping;
• Establish retroactivity under § 2255(f)(3);
• Allow the government to waive mandatory restitution;
• Invalidate every restitution order arising from pre-1996 conduct;
• Decide an Excessive Fines Clause challenge;
• Require consideration of ability to pay when calculating the amount; or
• Transform every civil monetary sanction into criminal punishment.

The decision answers one foundational question. The consequences of that answer will be developed incrementally through future cases.

The Bottom Line

Ellingburg v. United States marks an important constitutional shift.

The Supreme Court has rejected the fiction that MVRA restitution is merely a civil debt designed only to compensate victims. It is part of the criminal sentence, imposed by the government to punish a public wrong.

That holding directly matters when the government attempts to apply harsher restitution consequences retroactively to conduct predating the MVRA. It may also influence future debates over procedure, proportionality, financial sanctions, and the constitutional treatment of government-imposed monetary penalties.

But Ellingburg does not eliminate the separate custody requirement of § 2255.

Across the federal courts of appeals, the predominant rule remains that an ordinary restitution-only claim does not seek release from custody and therefore falls outside § 2255—even when the motion is timely and even when the petitioner simultaneously challenges a prison sentence.

That procedural limitation does not diminish the importance of Ellingburg. It exposes the central contradiction more clearly:

The federal system now recognizes restitution as criminal punishment while continuing to deny defendants the ordinary collateral remedy available for custodial punishment.

The next phase of the debate will concern whether courts and Congress are prepared to treat restitution consistently with what the Supreme Court has declared it to be—not an administrative afterthought, not simply a civil debt, but punishment.

---

Disclaimer

This article is provided for general educational and informational purposes only. It does not constitute legal advice, create an attorney-client relationship, or evaluate the procedural remedies, limitations periods, waiver issues, or governing precedent applicable to any individual case.

Selected Authorities

• Ellingburg v. United States, 607 U.S. 163 (2026).
• 18 U.S.C. §§ 3613, 3613A, 3614, 3663A, and 3664.
• 28 U.S.C. § 2255.
• Smullen v. United States, 94 F.3d 20 (1st Cir. 1996).
• Kaminski v. United States, 339 F.3d 84 (2d Cir. 2003).
• United States v. Rutigliano, 887 F.3d 98 (2d Cir. 2018).
• United States v. Hatten, 167 F.3d 884 (5th Cir. 1999).
• Weinberger v. United States, 268 F.3d 346 (6th Cir. 2001).
• Amaya v. United States, No. 21-1781 (6th Cir. June 23, 2023).
• United States v. Miclaus, No. 23-3146 (6th Cir. May 29, 2026).
• Barnickel v. United States, 113 F.3d 704 (7th Cir. 1997).
• United States v. Bernard, 351 F.3d 360 (8th Cir. 2003).
• United States v. Kramer, 195 F.3d 1129 (9th Cir. 1999).
• Blaik v. United States, 161 F.3d 1341 (11th Cir. 1998).
• Mamone v. United States, 559 F.3d 1209 (11th Cir. 2009).
• Ramdeo v. United States, 136 F.4th 1348 (11th Cir. 2025).