How do attorneys use disparity evidence in a sentencing argument?

The federal sentencing statute directs a court to consider "the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct" (18 U.S.C. § 3553(a)(6)). Attorneys read that factor as a footing to place a proposed sentence against the recorded distribution for the same charge and criminal-history profile, so the argument is grounded in the aggregate data rather than an assertion. The aggregate figures orient that argument; they do not decide it, and they do not measure any judge.

Does a higher district median prove that district's bench is more severe?

No, not by itself. A district's overall median reflects its case mix as much as its bench. The border districts post low medians largely because their dockets are heavy with high-volume, low-level immigration cases, not because their judges are lenient on serious crime. That is why the study also reports the median within a single charge type, where the geographic gap still holds. Read the figures as descriptive, not as a finding about any court or judge.

Where does the data come from and can it be reproduced?

The figures are a descriptive aggregate of the 378,458 FY2020–FY2025 individual-offender records in the public U.S. Sentencing Commission datafile, snapshot 2026-06-26. Each district figure is the median (50th percentile) of total prison months for records with a numeric prison term; probation-only and life-coded terms are excluded from the median. Every number is reproducible from the public source by re-running the published analysis script, and the analysis identifies no individual. The full breakdown by district and by charge is in the companion data study.

Guide · Federal sentencing

Federal sentencing disparity by district, and what the data actually shows.

The same federal offense does not carry the same typical sentence everywhere it is filed. Across the public U.S. Sentencing Commission data, the median prison term varies widely by judicial district, and the gap persists even after the charge is held constant. This is a working reference to what that disparity is, what the aggregate figures show, and how the data informs a sentencing argument under the statutory factor that addresses it, so the conversation starts from the records rather than an assertion.

Free · search by district and charge

See the median federal prison sentence for every judicial district, overall and within the four highest-volume offense categories, drawn from 378,458 public USSC individual-offender records (FY2020–FY2025), with a downloadable CSV. No account, no upload.

Open the federal sentencing-disparity data study →

Read the figures below as descriptive, not causal, and not predictive. The medians are recorded across many different cases. They are not adjusted for offense severity, criminal history, mandatory minimums, or acceptance of responsibility, they do not measure any court or judge, and they are not a prediction of what any defendant will receive. A district median reflects its case mix as much as its bench, so a higher or lower figure reflects what and who is in each group. This is not a ranking of judges by severity.

What inter-district disparity is

Federal sentences are imposed under one national code, yet the typical sentence for the same broad offense differs by where the case is filed. Across 378,458 individual-offender records in the public USSC datafile (FY2020–FY2025), the national median prison sentence is 21 months. Grouped by judicial district, the median ranges from 4 months in D. New Mexico to 87 months in S.D. Iowa, a 21.8× spread. That roughly 22-fold range is a descriptive summary of past records across the 94 federal districts, not a claim about any court's conduct.

What the aggregate USSC data shows

The overall district spread is the headline, but a single overall median is a blunt figure: much of it reflects each district's case mix. The border districts post low overall medians largely because their dockets are heavy with high-volume, low-level immigration cases, not because their benches are lenient on serious crime. That is why the study also reports the median within a single charge type, and the geographic gap persists there too.

For drug-trafficking cases, one of the two largest federal offense categories by count (n=107,244), the district median ranges from 12 mo to 132 mo across the 92 districts with at least 100 such cases, against a national median of 60 mo. Same charge, an order-of-magnitude difference in the typical exposure depending on where it is filed. Firearms, fraud, and immigration cases show the same pattern. Holding the charge constant isolates more, but not all, of the local charging-culture and judicial component; the remaining figures are still descriptive, not causal.

Figures reproduced from the federal sentencing-disparity data study, derived from the public USSC Commission individual-offender datafile (snapshot 2026-06-26; 378,458 records; the geographic table covers the 93 districts clearing a 200-record floor).

Even inside one state, the median diverges

Disparity is not only a red-state / blue-state or coastal / interior artifact. Florida's three federal districts, all in the same circuit, post different overall medians: M.D. Florida at 46 mo, N.D. Florida at 40 mo, and S.D. Florida at 30 mo. The same divergence appears within many states. It is one more reason the figures are read as descriptive orientation, tied to case mix and local practice, rather than as a measure of any single bench.

How the data informs a § 3553(a)(6) argument

The federal sentencing statute directs a court, in determining the sentence, to consider a list of factors. One of them, 18 U.S.C. § 3553(a)(6), is stated in the statute's own words as “the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct.” That is the statutory text, quoted verbatim, and nothing more is asserted about how any court has applied it.

Read against that factor, the aggregate distribution is a way to place a proposed sentence in the context of what similarly situated cases have actually received, so an argument about avoiding an unwarranted disparity rests on the recorded data rather than on a bare assertion. The aggregate orients that argument; it does not decide it. Whether a particular difference is warranted turns on the individual record, criminal history, mandatory minimums, and the offense-specific guideline calculation, none of which an aggregate median captures. Counsel using the data draws the comparable for the specific charge and history rather than citing the national spread as if it settled the question.

Statutory factor: 18 U.S.C. § 3553(a)(6) (the need to avoid unwarranted sentence disparities). Quoted as the statute's own text; no case holding is asserted here.

What the aggregate data does and does not tell you

The value of the aggregate is orientation: it tells you the order of magnitude and the shape of the sentencing distribution by district and by charge. It does not, and cannot, predict what a specific client will receive, which depends on the offense, the guideline range, criminal history, any mandatory minimum, acceptance of responsibility, and the court's judgment. A district's median reflects its case mix as much as its bench, so the figures do not measure any judge and are not a ranking. They are a descriptive picture of past public records, used to frame an argument, never to make a finding about a person or a court.

The data study breaks the median down by district and by offense category, so you can pull the recorded distribution for the specific charge in front of you, with a downloadable CSV and the full methodology.

Open the federal sentencing-disparity data study →

Reference

For the full ranking of district medians, the offense-controlled cuts, and the methodology, with a downloadable CSV and JSON, see the federal sentencing-disparity data study. For the full set of free federal sentencing studies, start at the federal sentencing data hub.

Common questions

What is inter-district federal sentencing disparity?: It is the variation in the sentence imposed for federal offenses depending on which of the 94 federal judicial districts hears the case. Across 378,458 individual-offender records in the public USSC datafile (FY2020–FY2025), the national median prison sentence is 21 months, but the district median ranges from 4 months in D. New Mexico to 87 months in S.D. Iowa, a 21.8× spread. That range is a descriptive summary of past records, not a prediction of any one case.
How do attorneys use disparity evidence in a sentencing argument?: The federal sentencing statute directs a court to consider "the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct" (18 U.S.C. § 3553(a)(6)). Attorneys read that factor as a footing to place a proposed sentence against the recorded distribution for the same charge and criminal-history profile, so the argument is grounded in the aggregate data rather than an assertion. The aggregate figures orient that argument; they do not decide it, and they do not measure any judge.
Does a higher district median prove that district's bench is more severe?: No, not by itself. A district's overall median reflects its case mix as much as its bench. The border districts post low medians largely because their dockets are heavy with high-volume, low-level immigration cases, not because their judges are lenient on serious crime. That is why the study also reports the median within a single charge type, where the geographic gap still holds. Read the figures as descriptive, not as a finding about any court or judge.
Does the geographic gap survive controlling for the offense?: Yes. For drug-trafficking cases, one of the two largest federal offense categories by count (n=107,244), the district median ranges from 12 to 132 months across the 92 districts with at least 100 such cases, against a national median of 60 months. Firearms, fraud, and immigration cases show the same pattern. Holding the charge constant isolates more, but not all, of the local charging-culture and judicial component.
Where does the data come from and can it be reproduced?: The figures are a descriptive aggregate of the 378,458 FY2020–FY2025 individual-offender records in the public U.S. Sentencing Commission datafile, snapshot 2026-06-26. Each district figure is the median (50th percentile) of total prison months for records with a numeric prison term; probation-only and life-coded terms are excluded from the median. Every number is reproducible from the public source by re-running the published analysis script, and the analysis identifies no individual. The full breakdown by district and by charge is in the companion data study.

This guide is general legal-information for practitioners, not legal advice, and the figures are descriptive aggregates over past public records, not a prediction of any specific case. Confirm every figure against the underlying data and the applicable statute and guideline before relying on it.