Abstract

Excerpted From: Eric S. Fish, Sentencing Immigrants, 99 Southern California Law Review 291 (December 2025) (447 Footnotes) (Full Document).

The most commonly charged federal felony is the crime of reentering the United States after deportation. On average twenty thousand people have been prosecuted for it per year since 2010, making unlawful reentry cases about one-third of the federal criminal caseload. And that number is about to dramatically increase. President Trump’s Department of Justice has announced a new policy of charging felony unlawful reentry in every available case, and it plans to reassign federal prosecutors working on drug and terrorism cases in order to prosecute more immigrants. Roughly 99% of the defendants in unlawful reentry cases are Latin American. The charge is, by definition, a nonviolent and victimless crime. It involves only entering or being found in the United States after a deportation. Yet defendants in these cases are systematically punished more harshly than comparable U.S. citizen defendants are in other cases. This happens in two ways.

First, the United States Sentencing Guidelines count reentry defendants’ criminal history against them twice. The Sentencing Guidelines are a complex set of rules that produce recommended sentences in federal criminal cases. While federal judges are not bound by the Guidelines, they follow the Guidelines about 90% of the time in unlawful reentry sentencings. In a standard criminal case, the Guidelines create a recommended sentence range by quantifying two variables: the defendant’s past convictions and facts about the current case. Past convictions create a “Criminal History Category” between one and six, while facts about the current crime create an “Offense Level” between one and forty-three. But in unlawful reentry cases, past convictions increase both the Criminal History Category and the Offense Level. Reentry defendants thus receive much larger sentence enhancements for their past crimes.

This double counting is worsened by another problem: the arbitrary way prior convictions are used to generate enhancements. The unlawful reentry guidelines make large swings in sentence length turn on obscure details of a person’s criminal history. This renders sentencing in these cases uniquely random and uncertain. Multiple years in prison turn on minutiae like the precise timing of a deportation order or probation violation, or whether a state uses a determinate or indeterminate sentencing system. People with similar criminal and immigration histories receive vastly different sentences based on such technical differences. And this problem is further compounded by the difficulty of finding and interpreting state conviction records. Figuring out the recommended sentence is like solving a high-stakes logic game. Defense lawyers are commonly unable to discern what length of sentence a defendant is looking at. And because the system pressures reentry defendants to plead guilty at the first opportunity, defendants often find out at sentencing that their punishment will be more severe than anticipated.

Second, undocumented immigrants serve a substantially larger portion of their sentences than do U.S. citizen defendants. The First Step Act, signed into law by President Trump in 2018, created a new system of “earned time” credits for federal prisoners. Under that law, a federal defendant who is willing to participate in anti-recidivism programs can get up to fifteen days off their prison sentence for every thirty days served. That means their time in prison can be reduced by up to one-third. In addition, defendants receive these credits even if they are not assigned to a program or no program is available. The only requirement is that they not refuse to participate. But deportable immigrants are prohibited from both participating in First Step Act programs and receiving custody reductions. This means that, by law, an undocumented prisoner will serve a significantly higher percentage of their sentence due to their immigration status.

To illustrate how this sentencing discrimination works in practice, consider two federal defendants: X and Y. Defendant X is charged with escaping from a federal prison, while defendant Y is charged with reentering the United States after deportation. Both X and Y have identical criminal histories: two prior felony convictions within the past fifteen years, each resulting in a two-year sentence. Under the Sentencing Guidelines, both X and Y will have a Criminal History Category of III due to the two prior felonies. Assuming that there are no aggravating or mitigating facts (like using physical violence or returning to custody voluntarily), X will have an Offense Level of thirteen for escaping from prison. Y will start with a lower Offense Level of eight for reentering the United States. But Y’s past convictions will also trigger enhancements of the Offense Level. If Y’s first deportation happened either before or after their two felony convictions, their total Offense Level will be sixteen. If their first deportation happened in between the two felony convictions, the Offense Level will be twenty-four. The Guidelines will recommend a sentence of eighteen to twenty-four months for X’s prison escape. And for Y’s reentry, they will recommend a sentence of either twenty-seven to thirty-three months or sixty-three to seventy-eight months (depending entirely on the timing of the first deportation). Unlawful reentry is the only federal crime for which criminal history is double counted like this. And once in prison, X will qualify for First Step Act programs that will let X spend up to one-third of their sentence out of custody or at a halfway house. Y, as a deported immigrant, cannot benefit from those programs.

As this example shows, these rules create two separate and unequal federal sentencing systems. Undocumented defendants, basically all of them Latin American, are treated more severely for their past crimes and are denied the lenient sentence reductions that U.S. citizens enjoy. And the enactment history of these policies reflects nativist animosity toward Latin American immigrants. The large recidivist enhancement for unlawful reentry was first created in 1991 with no study and little debate. It was enacted in response to Congress increasing the maximum sentence from two to fifteen years. That statutory change, in turn, was driven by racial panic in Florida over Haitian and Cuban refugees causing a supposed crime wave. The initial enhancement focused only on a small number of serious crimes, including murder and drug trafficking. But it evolved over time into a general enhancement for all past crimes. The First Step Act, in turn, was the first Trump Administration’s signature criminal justice reform initiative. Since the Trump Administration has been the most anti-immigration presidency in recent memory, it is not surprising that the law excludes deportable immigrants from its earned time credit system.

These sentencing rules explicitly discriminate against immigrants, and their history reflects racist and nativist antipathies. It might thus be logical to look to the Equal Protection Clause for a remedy, or to the federal statute mandating that the Sentencing Guidelines be “entirely neutral” as to race and national origin. But defendants trying to make such arguments have run into a problem--the Supreme Court’s antidiscrimination doctrines erect numerous obstacles to such equality claims. Federal laws that discriminate by immigration status are only subjected to “rational basis” review. While these sentencing rules overwhelmingly burden Latin American defendants, the Supreme Court has disallowed racial discrimination claims that rely on disparate impact. And the Court adopts a strong presumption that government action is not motivated by racial animus.

But sentencing is not like judicial review. In the discretionary sentencing context, the judge is the legally designated decisionmaker. They are not reviewing the work of an agency or legislature. Federal judges are empowered to disregard the Guidelines and sentence a defendant anywhere within the statutory range. The deference concerns that have hollowed out Equal Protection doctrine thus do not apply to discretionary sentencing decisions. In other contexts, scholars have identified circumstances in which executive branch agencies engage in “administrative constitutionalism”--incorporating constitutional norms into policy more robustly than formal legal doctrine requires. Judges should also embrace a version of administrative constitutionalism at sentencing. Here I call this ““sentencing constitutionalism.” Employing it, judges should reject sentencing rules that create racial and immigration-status-based hierarchies. And they should do so even if those rules would be upheld in the formal constitutional review context. Justice Kennedy gestured at something like this approach in Beckles v. United States. In that case, the Supreme Court held that constitutional vagueness challenges do not apply to the Federal Sentencing Guidelines because they are advisory. But Justice Kennedy wrote a concurrence arguing that quasi-constitutional vagueness arguments still have a place in the discretionary sentencing context. Sentencing constitutionalism is also analogous to the idea of “imperfect” defenses in criminal cases. There are contexts in which a defendant cannot legally argue a defense like duress or self-defense, but the moral justification underlying that defense still clearly applies. Judges will often reduce defendants’ sentences due to such imperfect defenses. Sentencing constitutionalism operates the same way, but with constitutional equality claims in place of defenses.

Federal crack cocaine sentencing provides a powerful example of sentencing constitutionalism in practice, as well as a close analogue to the sentencing discrimination faced by immigrants. For sentencing purposes, the Guidelines initially treated one gram of crack cocaine the same as one hundred grams of powder cocaine. This caused crack cocaine traffickers to be punished much more harshly than powder cocaine traffickers for the same volume of basically equivalent drugs. And this had a clear disparate impact by race because crack cocaine defendants were almost entirely African American. Much like the policy of double-counting immigrants’ criminal history, these crack sentencing rules were designed without formal deliberation or policy analysis, and were made in response to a congressional statute enacted during a racialized moral panic over crime. The crack sentencing guidelines were ultimately upheld in direct constitutional challenges. But many federal judges have exercised their authority to disregard the crack guidelines and adopt less discriminatory sentencing practices. In doing so, these judges have adhered to a higher standard of antidiscrimination than formal doctrine requires. Judges should act similarly in the immigrant sentencing context. They should refuse to impose the discriminatory rules built into the Guidelines, and they should counteract the discrimination built into the First Step Act’s sentence reductions.

This Article has two goals: to describe how federal sentencing law discriminates against undocumented immigrants, and to propose a framework for judges to counteract that discrimination. It is organized into four Parts. Part I traces the history of federal immigrant sentencing over the last several decades, exploring how and why these unequal rules were adopted. Part II shows that these sentencing rules discriminate explicitly by immigration status, and in effect by race. Part III explains how the Guidelines’ double-enhancements for criminal history create a sentencing system where arbitrary details of a defendant’s criminal record dictate large (often multi-year) increases in their sentence. It also illustrates the problem with a real-life case. Part IV introduces the concept of sentencing constitutionalism. It argues that judges should exercise their inherent power over discretionary sentencing to counteract punishment rules that discriminate by race and immigration status. It also shows that federal judges have been doing precisely this in the context of crack cocaine sentences--varying downward to mitigate racial discrimination that has gone unredressed in formal legal challenges.

[ . . . ]

This Article has shown that federal sentencing laws explicitly discriminate against undocumented immigrants. The Sentencing Guidelines give them duplicative recidivism enhancements that do not apply to U.S. citizens. The First Step Act denies them sentence reductions that U.S. citizens receive. This discrimination contributes to a racial and nationality-based hierarchy in the federal criminal justice system. Undocumented immigrants, nearly all of them Latin American, prosecuted for entering the United States, are treated as a legal underclass. They are punished more severely for past crimes and granted less mercy for current rehabilitation. Fortunately, federal district judges can help counteract this discrimination. Sentencing judges have the power to disregard the Guidelines’ duplicative enhancements. They also have the power to correct for immigrants’ exclusion from in-custody sentence reduction programs. For nearly twenty years, judges have mitigated a similar injustice in crack cocaine cases by systematically lowering sentences. They have done so out of recognition that the sentences for crack cocaine are racially discriminatory, and that this discrimination has gone unremedied in ordinary litigation. Judges should do the same for undocumented immigrants.

Professor of Law, University of California at Davis.