Plea Agreements: Federal Court vs. Courts-Martial, 2016–2025
In fiscal years 2016 to 2025, 472,754 of 642,522 people sentenced in federal court after a guilty plea (74%) had a written plea agreement. In courts-martial with a guilty plea and a plea-agreement entry, 4,020 of 4,118 (98%) had a plea agreement. The numbers come from separate record systems, and Ai
Understanding Trends in Military Practice · 6 min read
Trends › Federal Prosecution Analysis › Guilty Pleas in Federal Court vs. Courts-Martial, Year by Ye › Plea Agreements
In brief: In fiscal years 2016 to 2025, 472,754 of 642,522 people sentenced in federal court after a guilty plea (74%) had a written plea agreement. In courts-martial with a guilty plea and a plea-agreement entry, 4,020 of 4,118 (98%) had a plea agreement. The numbers come from separate record systems, and Air Force records do not record plea agreements at all.
This post uses the U.S. Sentencing Commission's records of 659,036 people sentenced in federal court in fiscal years 2016 to 2025, compared with court-martial results from this site's Military Trial Results data (updated November 2, 2026). Acquittals and dismissals come from a second source, the Federal Judicial Center's records of 753,808 federal defendants whose cases ended in fiscal years 2016 to 2025. It is part of the Federal Prosecution Analysis series, which builds on Military Moves Toward Sentencing Guidelines to Mirror the Federal System. How the numbers were built: How the Federal Comparison Data Was Built.
How Common Were Written Plea Agreements in Federal Court?
The U.S. Sentencing Commission recorded 642,522 people who pleaded guilty from fiscal year 2016 through fiscal year 2025. Of those, 472,754 (74%) had a written plea agreement. Another 2,807 (less than 1%) had an oral agreement.
The remaining 166,518 (26%) pleaded guilty with no agreement. A guilty plea without an agreement is sometimes called a straight plea. The person admits the charge, and the judge decides the sentence without any negotiated terms.
How Did the Federal Share Change by Year?
The share with a written agreement was not constant. It was 74% in fiscal year 2016 and 68% in fiscal year 2025, which is the lowest of the years in the table. The highest was 79% in fiscal year 2022.
The number of guilty pleas also varied. The table shows 74,709 in fiscal year 2019 and 56,324 in fiscal year 2021. The records do not show why the share or the count changed in any year.
| Fiscal year | Federal guilty pleas | Written agreement | No agreement |
|---|---|---|---|
| 2016 | 65,893 | 74% | 25% |
| 2017 | 64,981 | 74% | 25% |
| 2018 | 67,610 | 71% | 28% |
| 2019 | 74,709 | 70% | 29% |
| 2020 | 63,157 | 69% | 30% |
| 2021 | 56,324 | 78% | 22% |
| 2022 | 62,527 | 79% | 20% |
| 2023 | 62,300 | 77% | 22% |
| 2024 | 59,922 | 76% | 24% |
| 2025 | 65,099 | 68% | 32% |
How Common Were Plea Agreements in Courts-Martial?
For courts-martial with trial dates from 2016 to 2025, 4,118 cases had a guilty plea and a plea-agreement entry. Of those, 4,020 (98%) had a plea agreement.
The Army had 1,577 of 1,577. The Navy had 1,270 of 1,295, and the Marine Corps had 1,173 of 1,246. The Army figure means every Army guilty plea with an entry showed an agreement.
Air Force records in this data do not record plea agreements. No Air Force figure is given here, and the 98% does not include that service.
Do the Percentages Measure the Same Thing?
They do not count the same group. The federal figure counts everyone sentenced after a guilty plea. The court-martial figure counts only guilty pleas that have a plea-agreement entry, so cases with no entry are left out.
The federal data also contains only people who were convicted and sentenced. Acquittals and dismissals are not in it. A reader should treat 74% and 98% as separate measurements, not as a rate that can be compared line by line.
What Does a Plea Agreement Control?
Since January 1, 2019, a military plea agreement can set a minimum and maximum sentence. The whole agreement is shown to the military judge, and the judge sentences within its limits once the judge accepts it.
Most federal plea agreements do not bind the judge on the sentence. An agreement under Federal Rule of Criminal Procedure 11(c)(1)(C) does bind the judge once the judge accepts it. So an agreement can mean different things in federal court and in courts-martial, even when both are counted as an agreement.
What Do These Records Not Show?
The records do not show what each agreement contained, such as charges dropped, a sentence cap, or a promise to cooperate. They do not show how many people were offered an agreement and refused it. They also do not show why the federal share was lower in some years than in others.
A count of agreements says how often a written deal existed. It does not say whether the sentence was lower than it would have been without one.
How We Counted
Federal: people sentenced after a guilty plea in fiscal years 2016 to 2025 (642,522, U.S. Sentencing Commission, plea agreement field DSPLEA). Courts-martial: guilty pleas with a plea-agreement entry (4,118).
Federal fiscal years run from October 1 to September 30. No person, judge, or lawyer is identified in any of these records.
What to Keep in Mind
Who is in each set of numbers. The federal sentencing numbers come from the U.S. Sentencing Commission. They count only people who were convicted and sentenced in federal district court. People whose cases were dismissed or who were acquitted are not in them. The military numbers come from court-martial results, which include acquittals. To compare acquittal or dismissal rates, the federal side uses a different source: the Federal Judicial Center's case data, which counts every defendant whose case ended.
Plea agreements work differently. In federal court, most plea agreements do not bind the judge on the sentence. Since 2019, a military plea agreement can set a minimum and maximum sentence that the military judge must follow if the judge accepts the agreement. The federal numbers count anyone who pleaded guilty, with or without a written agreement.
How "plea" and "trial" are counted. In the federal sentencing numbers, "trial" means the person was convicted after a trial, including people who pleaded guilty to some counts and went to trial on others. People acquitted at trial are not in these numbers at all (see "Who is in each set of numbers"). In the federal case data, the trial rate counts both convictions and acquittals at trial among all defendants whose cases ended, including dismissed cases. Military numbers count contested cases that reached a verdict. These denominators are not the same, and each post states the one it uses.
Time periods. Federal numbers are by federal fiscal year (October 1 to September 30), based on the sentencing date. The military numbers are by trial date. Neither source has the date of the offense, so a change in law shows up in either data set gradually.
Related Analysis
- Military Moves Toward Sentencing Guidelines to Mirror the Federal System (the background article)
- Federal Prosecution Analysis (all federal posts)
- How the Federal Comparison Data Was Built (sources and method)
- Guilty Pleas in Federal Court vs. Courts-Martial, Year by Year (overview)
- U.S. Sentencing Commission, Individual Offender Datafiles (source)
- Federal Judicial Center, Integrated Database (criminal defendants) (source)
About the author. Jocelyn C. Stewart is a retired Army JAG with more than 24 years of trying courts-martial and more than 200 jury trials. She hosts Where All The Bodies Are Buried and runs the Law Office of Jocelyn C. Stewart. This analysis is for education and information only. It is not legal advice, and every case is different.