What the datafile does and does not include
What is in it
Federal courts send the Commission the sentencing documents for every felony and Class A misdemeanour case: the judgment, the presentence report, the statement of reasons and the plea agreement. The Commission codes each case into its individual offender datafile for the fiscal year in which sentence was imposed. Over FY2018–FY2025 that is 524,421 cases. The file carries no names; cases are identified only by a Commission serial number, which this site does not store.
What is not in it
- Petty offences and most Class B and C misdemeanours.
- Cases sentenced in the year but whose documents reached the Commission late (they appear in a later file, or not at all).
- Organisational defendants, which have a separate datafile.
- Judges. The Commission does not release the identity of the sentencing judge, and this site publishes nothing about individual judges.
- Anything after sentencing: appeals, resentencing, compassionate release and supervised-release violations are not reflected.
How sentences are recorded
- Life and any term of 470 months or longer is recorded as 470 months in the capped variable this site uses, so that a handful of very long terms do not dominate averages. 2,286 cases over the period are at the cap.
- Time served is counted as prison time when its length is known; when the documents say only "time served" the Commission records one day.
- Probation and alternatives without prison have no prison-months value. They are excluded from medians and the histogram, and reported as the share of cases without a prison term (8% nationally).
- Alternative confinement (home detention, community confinement) served alongside prison is not added to prison months.
Fiscal years
The federal fiscal year runs from 1 October to 30 September, so FY2025 covers 1 October 2024 to 30 September 2025. The Commission publishes each year's file the following spring and re-releases earlier years alongside it.
Why this site starts in FY2018
The Commission changed its offence-type classification, its range-status categories and its sentence-length conventions with the FY2018 file. Earlier years can be analysed with the older variables, but mixing the two would make trends unreliable, so the site begins where the current definitions begin.