Blog
Federal Sentencing
The guidelines do not sentence anyone. A judge does, in a courtroom where the arithmetic of offense levels and criminal history categories has already been performed by a probation officer the defendant did not choose and may never see again. The question is what the judge decides to do with that recommendation, and how prepared the person standing at the lectern has been for the moment when a number becomes a duration.
The Two-Step Process
Before November 2025, federal sentencing followed three steps: calculate the guideline range, consider departures, then evaluate the § 3553(a) factors for a possible variance. In fiscal year 2024, courts applied variances in roughly a third of sentencing proceedings. They applied departures, including government-sponsored departures, in approximately four percent. The Commission, recognizing what practitioners had understood for some time, collapsed the second and third steps into one.
The word “departure” has been removed from the operative text of the Guidelines Manual. The provisions that once governed departures now reside in Appendix B, preserved for historical reference. Two exceptions survived the consolidation: § 5K1.1, which permits a reduced sentence when a defendant has provided substantial assistance to law enforcement, and early disposition programs under the relocated § 3F1.1. Everything else flows through § 3553(a).
In practice, this changes less than it appears to change. Judges in most districts were already bypassing the departure analysis and proceeding directly to the statutory factors. The 2025 amendments formalized what had become custom.
But formalization has consequences. A sentencing memo that argued for a departure under the old framework spoke a different language than one arguing for a variance under § 3553(a). The departure framework was mechanical: identify the provision, satisfy its criteria, apply the reduction. The variance framework is narrative. It requires the attorney to construct an argument about who the defendant is, what the offense represents in the full context of a life, and why the guideline range produces a sentence greater than necessary.
Whether every federal defender in the country has absorbed this shift is, if we are being precise, not a question with an obvious answer.
The Presentence Report
The presentence investigation report is the document that determines more than any sentencing memorandum and receives the least attention from defendants at the moment it requires the most. Probation officers compile it after conviction, drawing from interviews, criminal records, financial disclosures, and the government’s version of the offense. The report calculates the guideline range. It also contains a narrative of the offense conduct that the judge will read before the sentencing hearing, and that narrative is not the defendant’s.
Objections to the PSR must be filed within a window that varies by district but is rarely generous. In some districts, the defense has fourteen days. The objections that matter are the ones that challenge the factual basis for specific offense characteristics: the drug quantity attributed, the loss amount calculated, the role enhancement applied. A two-level difference in offense level can translate to years.
Most defendants do not understand this until the report arrives, and the report arrives on its own schedule. By then, the calculation is done and the narrative belongs to the probation officer.
The Weight of § 3553(a)
Section 3553(a) of Title 18 instructs that a sentence shall be sufficient, but not greater than necessary, to serve the purposes of sentencing. The statute identifies those purposes as just punishment, deterrence, protection of the public, and rehabilitation where appropriate. The factors include the nature and circumstances of the offense, the history and characteristics of the defendant, the guideline range itself, and the need to avoid unwarranted sentencing disparities among defendants with similar records found guilty of similar conduct.
After Booker rendered the guidelines advisory in 2005, and after Gall v. United States clarified in 2007 that appellate courts could not presume below-range sentences unreasonable, the § 3553(a) factors became the center of federal sentencing advocacy. Federal sentencing has always been a negotiation between a formula and a person. The question is how much of the person the court is permitted to see.
There is a kind of sentencing memo that fails. It recites the factors, attaches letters from family members, and asks the court for mercy. The kind that succeeds tells a story the judge has not heard from the government, connects the defendant’s history to the offense in a way that makes the guideline range seem mechanical rather than just, and identifies a specific sentence that satisfies each statutory factor without requiring the court to perform the analysis alone. The memo does the judge’s work for the judge. The court adopts the reasoning, or does not, but the framework is there.
We approach § 3553(a) memoranda as litigation documents, not letters. The factual record that supports a variance begins before the PSR is drafted, not after. Mitigation specialists, where the case warrants one, conduct independent interviews and produce a social history that the probation officer’s report will not contain. The sentencing memorandum is built on that record. It does not ask the court to be lenient. It demonstrates, with specificity, that the guideline range overrepresents the seriousness of the offense, the criminal history, or both, and it proposes a sentence that accounts for what the guidelines were not designed to measure.
Whether this changes the outcome in every case is not a claim I would make. The judge who has already decided on a within-range sentence will impose it regardless of the memorandum’s quality. But the judge who senses that the arithmetic does not capture the case (and there are more of these judges than the government’s sentencing statistics suggest) needs a document that gives that instinct a legal structure. That is the document we compose.
The variance rate has increased since Booker. In the year before the decision, sentences within the guideline range accounted for more than seventy percent of federal sentences. By fiscal year 2024, courts applied variances in approximately thirty-two percent of proceedings. The trend is not uniform across districts. A sentencing in the Southern District of New York proceeds in a different climate than one in the Western District of Texas, and the climate matters as much as the statute.
The 2026 Proposed Amendments
The Sentencing Commission published proposed amendments for the 2026 cycle in two phases, December 2025 and January 2026. If Congress does not reject them, they take effect November 1, 2026.
The most consequential proposals concern economic crimes under § 2B1.1. The current loss table contains sixteen tiers. The proposed amendment collapses these to eight. For defendants facing fraud, embezzlement, or theft charges, the restructured table could lower guideline ranges, particularly at the middle tiers where the current system compresses conduct of substantially different severity into the same offense level. The DOJ has opposed the direction of these changes, arguing that reduced guidelines send a signal inconsistent with the harm economic crime continues to cause.
The Commission has also proposed narrowing the sophisticated means enhancement, a two-level increase that courts have applied with notable inconsistency. The proposed amendment would limit the definition to conduct exceeding the complexity typical for that type of offense, which, if adopted, would exclude schemes courts have treated as sophisticated simply because they involved a computer or a second bank account.
A new provision under § 3E1.2 would create a formal reduction for post-offense rehabilitation. Defendants who demonstrate sustained behavioral change between conviction and sentencing could receive a guideline credit. This codifies what defense attorneys have advised for as long as judges have held discretion, though the formal mechanism introduces questions about documentation, verification, and whether rehabilitation programs will be constructed to satisfy a checklist rather than to accomplish anything. Whether a formal credit changes behavior or merely incentivizes its performance is a question the Commission has not resolved and may not be positioned to resolve.
I am less certain about how the post-offense rehabilitation credit will function in practice than the Commission’s proposal suggests I should be. The concept is sound. The implementation depends on probation offices, defense counsel, and treatment providers agreeing on what sustained means, and that agreement does not exist yet.
The methamphetamine provisions are also under review. The current guidelines punish pure methamphetamine ten times more harshly than methamphetamine mixture (a distinction that, when adopted, served as a rough proxy for a defendant’s position in the distribution chain, on the theory that only high-level distributors handled the pure substance). Methamphetamine purity is now uniformly high across the market. The disparity now produces sentences driven by testing practices that vary from district to district rather than by anything resembling culpability.
Supervised Release and the Individualized Assessment
The 2025 amendments altered supervised release in ways that receive less attention than the sentencing changes but affect more people. More than a hundred thousand individuals are on federal supervised release at any given time. The prior framework defaulted to the statutory maximum term for most offenses and imposed conditions with minimal case-by-case differentiation. The revised § 5D1.1 requires an individualized assessment before supervised release is imposed. The court must explain its reasoning on the record.
Revocation proceedings have changed as well. The old system pushed toward automatic revocation for most violations. The new framework grades them: revocation is generally appropriate for Grade A violations (violence, firearms, controlled substances); often appropriate for Grade B violations (other felonies); and possible, though not presumed, for Grade C violations, which include misdemeanors and technical noncompliance.
For someone on supervision who misses a reporting date, or who tests positive for a substance legal under state law in the state where they reside, the difference between a system that defaults to revocation and one that permits a measured response is not a technicality. The revised framework does not guarantee the outcome the defendant wants. It permits a conversation the old framework did not, and permission is where most of the law’s actual work occurs.
The Commission is adjusting mechanisms, not issuing declarations.
What Remains
Federal sentencing is a system that operates on arithmetic and survives on judgment. The guidelines provide the architecture. The § 3553(a) factors provide the room. What a defendant receives depends, in the end, on the quality of the advocacy that fills that room before the judge speaks, and on whether the record before the court contains enough of the defendant’s life to make the number feel insufficient.
A consultation is where that work begins. A first conversation about sentencing costs nothing and assumes nothing; it is a diagnosis of where the numbers stand and what the record might support.