BenchRecon Data Study · Federal
A mandatory minimum, in months.
Across 378,458 federal sentencing records (USSC, FY2020–FY2025), the median prison sentence is 12 months with no mandatory minimum and 120 months when one applies, a 10× difference. But 35.3% of those convicted of a mandatory-minimum offense are relieved of it, which cuts the median to 46 months.
We analyzed every federal individual-offender record in the U.S. Sentencing Commission's public datafile for FY2020 through FY2025, 378,458 cases, and compared the median total prison sentence across three groups: no mandatory minimum, a mandatory minimum that was relieved (safety valve or substantial assistance), and a mandatory minimum the defendant remained subject to.
Read the gap carefully. Much of the raw 10× is selection, not the statute's own mechanism: the charges that carry a mandatory minimum are systematically more serious, a triggering drug quantity, a firearm, a serious prior record, all of which lengthen a sentence regardless. The raw 10× is an upper bound on what the minimum itself does. The charge-controlled table below holds the offense fixed and shrinks the gap sharply (drug trafficking 36 mo→120 mo, a 3.3× gap), closer to the statute's real effect, though still not a controlled experiment.
Aggregate analysis. Reproducible from public USSC data. No individual is named.
Who escapes the minimum
A mandatory minimum is not always the floor. Two statutory routes relieve it. The safety valve (18 U.S.C. §3553(f)) lets the court sentence below the minimum for lower-level, largely non-violent drug defendants with limited criminal history. A substantial-assistance motion (§5K1.1) does the same in exchange for cooperation. Of the 91,669 individuals convicted of a mandatory-minimum offense in this window, 64.7% remained subject to the minimum and 35.3% were relieved of it, a pooled-window split consistent with the U.S. Sentencing Commission's own published Quick Facts figure for FY2024 (37.0%/63.0%); the single-year share moves year to year, so a multi-year average lands near it rather than exactly on it.
The difference relief makes is large. The median for a defendant who remained subject is 120 mo; for a defendant who was relieved it is 46 mo, and the relieved group still sits well above the 12 mo median for cases with no minimum at all, because they were charged with serious conduct in the first place.
| Mandatory-minimum status | Median | Incarcerated | n |
|---|---|---|---|
| No mandatory minimum | 12 mo | 63.9% | 274,927 |
| Minimum relieved (safety valve / §5K1.1) | 46 mo | 85.2% | 32,364 |
| Remained subject to the minimum | 120 mo | 99.4% | 59,305 |
“Incarcerated” is the share with any prison time imposed (the rest receive probation or time served). Medians include those zero-month dispositions. The relieved group pools the safety-valve and substantial-assistance routes, matching how the Commission reports the category.
The gap shrinks, but holds, inside a single charge
The first objection to the raw 10× is that mandatory-minimum cases are simply different, more serious crimes. So we held the charge constant. Within drug trafficking, the largest mandatory-minimum category by volume of applied cases, the median is 36 mo with no minimum and 120 mo when one applies: a 3.3× within-offense gap, far below the raw 10× but still substantial. That collapse from 10× to 3.3× is the honest measure of how much of the headline was case-mix selection versus the statutory floor itself.
| Offense category | No minimum (n) | Relieved (n) | Subject (n) |
|---|---|---|---|
| Drug Trafficking | 36 mo (39,178) | 46 mo (29,681) | 120 mo (36,249) |
| Firearms | 36 mo (44,078) | 70 mo (433, thin) | 96 mo (4,040) |
| Fraud / Theft / Embezzlement | 6 mo (26,073) | 6 mo (409, thin) | 40 mo (2,685) |
| Immigration | 4 mo (120,676) | 18 mo (152, thin) | 36 mo (427, thin) |
Every cell carries its denominator (n). Drug trafficking is the only category with all three cells large; the relief cells for Firearms, Fraud, and Immigration are thin (n=152–433, flagged “thin”) and should be read as directional, not precise, mandatory-minimum relief outside drug cases is rare. Medians include probation/no-prison cases as zero months.
Methodology & limitations
Source & method
- Source: U.S. Sentencing Commission, Commission Datafiles, individual-offender records (public). ussc.gov/research/datafiles. Snapshot analyzed: 2026-06-26.
- Population: 378,458 FY2020–FY2025 individual-offender records, all with a numeric total prison term (USSC variable TOTPRISN, in months; probation/no-prison encoded as 0). The mandatory-minimum-convicted population is the 91,669 cases that either remained subject (n=59,305) or were relieved (n=32,364).
- Status coding: derived from the USSC variable MAND1. A defendant is classified as having no mandatory minimum, a relieved minimum (safety valve §3553(f) or substantial assistance §5K1.1), or remaining subject. The labels are validated two ways against the Commission's published Quick Facts (Mandatory Minimum Penalties, FY2024): our median sentence ordering (none 12 mo < relieved 46 mo < subject 120 mo) matches the Commission's mean ordering (31 < 70 < 157 mo), and our pooled-window relieved/subject split (35.3%/64.7%) is consistent with its FY2024 figure of 37.0%/63.0% (the single-year share varies by fiscal year, so a multi-year average lands near it, not on it).
- Statistic: the median (50th percentile, computed with
percentile_cont) of total prison months, grouped by status. The offense-controlled cuts restrict to a single USSC offense-guideline category (OFFGUIDE): Drug Trafficking, Firearms, Fraud/Theft/Embezzlement, and Immigration. - Reproducible: a published analysis script regenerates every figure on this page from the source data, and a companion verification script re-checks each number against the live datafile.
What the data does NOT show
- The raw 10× no-minimum-vs-applied gap is not a clean estimate of the mandatory minimum's own effect. It is heavily confounded by selection: a charge carries a mandatory minimum precisely because the conduct is serious, drug quantity, a weapon, a qualifying prior. Those raise the sentence independent of whether a minimum formally binds. The statute's mechanical effect is some fraction of the observed gap.
- Holding the offense constant removes the crudest confounder (different crimes) and shrinks the gap sharply, but it cannot control for case severity inside a single charge label: drug quantity, role, criminal history, and the specific count of conviction all vary within “drug trafficking.” This is an observational comparison, not a randomized one.
- Medians describe the typical case, not any individual sentence, and not the appropriate sentence in any matter. Eligibility for the safety valve and the decision to move for substantial assistance are themselves selective, so the relieved group is not a random subset of mandatory-minimum cases.
- This is the federal guidelines population only, not state-court sentencing. Do not infer anything about a specific judge, defendant, or case from these aggregate figures. This study identifies no individual.
Cite this analysis
Journalists and researchers, please link to this page as the source.
BenchRecon, “Federal mandatory minimums: a 10× jump in median prison time, and who escapes it” (USSC FY2020–FY2025 individual-offender data). https://benchrecon.com/data/federal-mandatory-minimums
The underlying figures come from the public USSC datafile and are reproducible from the published analysis script.
Quantify the exposure for your client
BenchRecon's Sentencing Snapshot turns this status-level signal into a case-specific comparables brief, the median and distribution for your charge, criminal-history category, and district, cited to the underlying USSC records, ready to file as a §3553(a) exhibit when you are arguing safety-valve eligibility or the weight of a mandatory minimum.
Common questions
- What does this study measure?
- The median total prison sentence, in months, for federal individual-offender cases grouped by mandatory-minimum status, FY2020–FY2025 (USSC data, n=378,458). With no mandatory minimum the median is 12 months (n=274,927); when a mandatory minimum applies and is not relieved the median is 120 months (n=59,305), a 10× difference.
- Is the whole 10× gap caused by the mandatory minimum itself?
- No, and the page does not claim that. A large part of the raw gap is selection, not the statute's mechanism. The charges that carry a mandatory minimum are systematically more serious: a triggering drug quantity, a firearm, a serious prior record. Those features lengthen a sentence regardless of whether a minimum formally binds. The honest reading is that the raw 10× is an upper bound on the statute's own effect, and the true mechanical effect is some smaller share of it. The charge-controlled tables below isolate more of that share.
- Then why is this still meaningful?
- Because the gap survives, in a smaller and more defensible form, when you hold the charge constant. Within drug trafficking, the largest mandatory-minimum category by volume of applied cases, the median is 36 mo with no minimum and 120 mo when one applies (n=36,249), a 3.3× within-offense gap rather than the raw 10×. Same broad charge type, far more time when a minimum binds, which is what a real statutory floor looks like once the crudest selection (different crimes) is removed.
- Who escapes a mandatory minimum, and how much does it matter?
- Two statutory routes relieve a mandatory minimum: the safety valve (18 U.S.C. §3553(f), for lower-level, largely non-violent drug defendants with limited criminal history) and a substantial-assistance motion (§5K1.1, for cooperation). Of the 91,669 individuals convicted of a mandatory-minimum offense in this window, 35.3% were relieved and 64.7% remained subject. Relief cuts the median sentence from 120 months (subject) to 46 months (relieved). That 35.3/64.7 pooled-window split is consistent with the U.S. Sentencing Commission's own published Quick Facts figure for FY2024 (37.0%/63.0%); the single-year share moves from year to year, so a multi-year average lands near, not exactly on, any one published year.
- Why use the median instead of the average?
- Sentence length is heavily right-skewed, a handful of very long terms (and life/very-high coded values) pulls the mean upward and misstates the typical case. The median (50th percentile) is the standard, robust summary of the typical sentence and is what the Sentencing Commission itself reports for distributions.
- Is this every federal case?
- It is the USSC individual-offender datafile for FY2020–FY2025, with a numeric total prison term (probation/no-prison encoded as 0 months and included in the medians). It covers original sentences under the guidelines; it is not a census of every disposition, and each cell shown carries its own record count (n).
- Can I cite or reuse this?
- Yes, please link to this page as the source. The underlying data is the public USSC datafile, and every figure is reproducible by re-running the published analysis script against it.