Compassionate Release Grant Rates by Circuit and District

By Elizabeth Franklin-Best, Esq., and Christopher Zoukis, JD, MBA

How Often Is Compassionate Release Granted?

Federal courts granted 14.1% of the compassionate release motions they decided in fiscal year 2025 — 394 grants out of 2,795 decisions (USSC FY2025 Compassionate Release Data Report, Tables 2–3). Through the first half of FY2026 the national rate was 14.0% (125 of 894). It depends enormously on the court: circuit rates ran from 4.9% to 31.0% in FY2025, and district rates from 0% to well over half.

Key takeaways on compassionate release grant rates

• The national rate has stayed in a narrow band across every period the Commission has published — 13.8% to 16.2%. It is not rising and it is not collapsing: 16.2% for FY2020–FY2022 combined (4,502 of 27,789), then 13.8% in FY2023 (432 of 3,140), 16.0% in FY2024 (481 of 3,015), 14.1% in FY2025 (394 of 2,795), and 14.0% through the first half of FY2026 (125 of 894).
• Circuit rates varied more than sixfold in FY2025 on identical law. The Ninth Circuit granted 31.0% (72 of 232); the Eighth granted 4.9% (26 of 528) (USSC FY2025 Data Report, Table 3).
• District spread is wider still. The District of Maryland granted 30 of 95 motions (31.6%) in FY2025. The Southern District of Iowa granted 0 of 81. Seven districts that decided ten or more motions granted none of them.
• You cannot choose your court. A § 3582(c)(1)(A) motion goes to the court that imposed the sentence, not the district where the person is imprisoned. Where the case was prosecuted years ago fixes which set of numbers applies to you.
• The motions that get granted are overwhelmingly the ones prisoners file themselves. Of the 394 grants in FY2025, 369 — 93.7% — were defendant-filed; the Bureau of Prisons Director filed 16 (Table 5).
• The sentencing factors under 18 U.S.C. § 3553(a) are the largest single reason courts give for denying. Courts cited them 1,260 times in FY2025 — 24.9% of the 5,067 reasons recorded across 2,401 denied cases. That is a share of reasons, not a share of cases.
• The Commission counts motions decided, not motions filed, and publishes no figure for the size of a sentence reduction. Both mistakes are common on other sites.

The National Grant Rate, Fiscal Year 2020 Through the First Half of Fiscal Year 2026

The Commission publishes a national total in Tables 2 and 3 of every report. Here is every period it has published, with the report each figure comes from.

Reporting period	Motions decided	Granted	Grant rate	Source report
FY2020–FY2022 combined (Oct. 1, 2019 – Sept. 30, 2022)	27,789	4,502	16.2%	FY2020–2022 report (Dec. 2022), Tables 1–3
FY2023	3,140	432	13.8%	FY2023 report (Mar. 2024), Tables 2–3
FY2024	3,015	481	16.0%	FY2024 report (Mar. 2025), Tables 2–3
FY2025	2,795	394	14.1%	FY2025 report (May 2026), Tables 2–3
FY2026, Q1–Q2 (Oct. 1, 2025 – Mar. 31, 2026)	894	125	14.0%	FY2026 Q2 report (May 2026), Tables 2–3

Why fiscal years 2020, 2021, and 2022 appear only as a three-year block

Because that is all the Commission published. For those three years it printed per-year granted and denied counts in table footnotes and a three-year aggregate — and nothing in between. The footnote counts are: FY2020, 1,819 granted and 5,471 denied; FY2021, 2,052 granted and 13,255 denied; FY2022, 631 granted and 4,561 denied (FY2020–2022 report, Tables 10–15 footnotes).

There is no published single-year total and no published single-year grant rate for FY2020, FY2021, or FY2022. Any percentage you see attributed to the Sentencing Commission for one of those years is somebody’s arithmetic wearing the Commission’s name. We decline to do that arithmetic here. What the aggregate does tell you is real: across the pandemic-era surge, courts decided 27,789 motions and granted 4,502 of them, 16.2%.

The counts move after publication, so the report matters as much as the number

The Commission’s data is limited to motions for which it has received court documentation and completed its analysis by a stated cutoff — April 27, 2026 for both the FY2025 and FY2026 Q2 reports. Documentation keeps arriving after that, so later reports carry higher counts for earlier years.

Summing the monthly rows of Table 1 by fiscal year — our arithmetic on the Commission’s published monthly counts, not a subtotal the Commission prints — the FY2025 report shows FY2023 at 442 grants out of 3,189 decisions, against the 432 out of 3,140 that the FY2023 report itself published. FY2024 shows as 485 of 3,047 against the originally published 481 of 3,015.

The rate barely moves; the counts do. That is why every figure on this page names the report it came from, and why you should never mix a figure from one report with a figure from another inside the same comparison.

Compassionate Release Grant Rates by Circuit

The Commission reports by circuit in Table 3. Twelve circuits appear — the eleven numbered regional circuits and the D.C. Circuit. The Federal Circuit is not among them: its jurisdiction is subject-matter specific — patents, government contracts, veterans’ claims and the like — and does not include criminal appeals from the district courts.

Circuit	FY2025 granted / decided	FY2025 rate	FY2026 Q1–Q2 granted / decided	FY2026 Q1–Q2 rate
Ninth	72 / 232	31.0%	21 / 79	26.6%
Second	32 / 118	27.1%	10 / 26	38.5%
First	14 / 56	25.0%	10 / 21	47.6%
D.C.	5 / 20	25.0%	1 / 3	33.3%
Tenth	21 / 101	20.8%	4 / 30	13.3%
Eleventh	75 / 447	16.8%	13 / 105	12.4%
Fourth	81 / 572	14.2%	39 / 196	19.9%
Third	12 / 95	12.6%	6 / 38	15.8%
Fifth	18 / 182	9.9%	6 / 63	9.5%
Sixth	24 / 271	8.9%	6 / 90	6.7%
Seventh	14 / 173	8.1%	3 / 56	5.4%
Eighth	26 / 528	4.9%	6 / 187	3.2%
NATIONAL	394 / 2,795	14.1%	125 / 894	14.0%

Source: USSC FY2025 Compassionate Release Data Report, Table 3, and USSC FY2026 Q1–Q2 report, Table 3. Sorted by FY2025 rate.

Read the half-year column with the N in front of you. The First Circuit’s 47.6% is 10 grants out of 21 decisions over six months, and the D.C. Circuit’s 33.3% is 1 out of 3. Those are not stable rates; they are small numbers that happen to divide well. The FY2025 column carries the denominators that can bear weight — 528 decisions in the Eighth Circuit, 572 in the Fourth, 447 in the Eleventh.

The stable finding across both periods is the ordering at the bottom. The Eighth, Seventh, Sixth, and Fifth Circuits each granted under 10% in FY2025 and again in the first half of FY2026, in the same order both times. If your sentencing court sits in one of those four, the base rate you are working against runs from 4.9% to 9.9%, against the Ninth Circuit’s 31.0%.

Compassionate Release Grant Rates by District

This is where the variation actually lives. The table below covers the districts that define the range — the busiest, the highest-granting, the zero-grant districts, the lowest nonzero rates at real volume, and all four federal districts of Texas. It is sorted by FY2025 motions decided, so the denominators you can trust sit at the top.

District	Circuit	FY2025 granted / decided	FY2025 rate	FY2026 Q1–Q2 granted / decided	FY2026 Q1–Q2 rate
S.D. Florida	11th	24 / 187	12.8%	4 / 37	10.8%
M.D. Florida	11th	14 / 147	9.5%	5 / 33	15.2%
E.D. Virginia	4th	17 / 135	12.6%	15 / 46	32.6%
W.D. Missouri	8th	4 / 114	3.5%	2 / 43	4.7%
E.D. North Carolina	4th	6 / 103	5.8%	0 / 26	0.0%
D. Maryland	4th	30 / 95	31.6%	9 / 31	29.0%
D. Minnesota	8th	9 / 83	10.8%	1 / 19	5.3%
S.D. Iowa	8th	0 / 81	0.0%	0 / 19	0.0%
N.D. Ohio	6th	2 / 80	2.5%	0 / 21	0.0%
D. South Carolina	4th	9 / 79	11.4%	7 / 33	21.2%
W.D. Texas	5th	4 / 62	6.5%	2 / 23	8.7%
S.D. Texas	5th	6 / 48	12.5%	4 / 16	25.0%
E.D. Kentucky	6th	1 / 38	2.6%	0 / 23	0.0%
D. Nebraska	8th	1 / 35	2.9%	0 / 18	0.0%
E.D. Wisconsin	7th	0 / 35	0.0%	0 / 9	0.0%
E.D. Texas	5th	1 / 34	2.9%	0 / 15	0.0%
E.D. Arkansas	8th	0 / 30	0.0%	0 / 20	0.0%
D. North Dakota	8th	0 / 28	0.0%	1 / 6	16.7%
D. Montana	9th	14 / 25	56.0%	4 / 7	57.1%
S.D. California	9th	13 / 17	76.5%	5 / 5	100.0%
N.D. Georgia	11th	9 / 16	56.3%	2 / 6	33.3%
N.D. Mississippi	5th	0 / 15	0.0%	0 / 3	0.0%
S.D. West Virginia	4th	0 / 15	0.0%	1 / 2	50.0%
N.D. Florida	11th	11 / 14	78.6%	1 / 1	100.0%
M.D. Louisiana	5th	0 / 11	0.0%	0 / 4	0.0%
S.D. Alabama	11th	5 / 6	83.3%	0 / 1	0.0%
N.D. Texas	5th	1 / 3	see note	0 decided	—
NATIONAL	—	394 / 2,795	14.1%	125 / 894	14.0%

Source: USSC FY2025 Compassionate Release Data Report, Tables 2–3, and USSC FY2026 Q1–Q2 report, Tables 2–3.

Note on the Northern District of Texas. One grant out of three motions decided is arithmetically 33.3%, and printing it that way would be misleading, so we do not. The number that means something in that row is the volume: the district decided 758 compassionate release motions across FY2020–FY2022, then 53 in FY2023, 13 in FY2024, 3 in FY2025, and none in the first half of FY2026. That is a docket that emptied out, not a court that became generous. The full four-district treatment is on our compassionate release texas page.

The same caution applies to the bottom third of the table generally. Once a district’s annual decision count drops under about twenty, a single ruling swings the rate by five or ten points. S.D. Alabama’s 83.3% is 5 grants out of 6 motions. S.D. California’s 100.0% for the first half of FY2026 is 5 out of 5. These are worth knowing and worthless for prediction.

The districts that granted nothing

Seven districts granted zero compassionate release motions in FY2025 while deciding ten or more: S.D. Iowa (0 of 81), E.D. Wisconsin (0 of 35), E.D. Arkansas (0 of 30), D. North Dakota (0 of 28), N.D. Mississippi (0 of 15), S.D. West Virginia (0 of 15), and M.D. Louisiana (0 of 11). The Southern District of Iowa is the one that stands out: 81 decisions, no grants, in a year when the national rate was 14.1%. Five of the seven sit in the Eighth Circuit or the Fifth.

Zero is not a bar. It is a base rate, and base rates are made of individual rulings on individual records. But it tells you what a motion is up against, and it should change how much documentation goes into the filing rather than whether the filing happens.

The districts that granted most often at real volume

D. Maryland is the standout: 30 grants out of 95 motions decided, 31.6%, against a 14.1% national rate — and it held at 29.0% (9 of 31) through the first half of FY2026. In FY2025 it was the only district in the country that both decided more than fifty of these motions and granted more than 30% of them. D. Montana granted 14 of 25 (56.0%), then 4 of 7 (57.1%). Below that, the high rates come with small denominators: N.D. Florida 11 of 14, S.D. California 13 of 17, N.D. Georgia 9 of 16, S.D. Alabama 5 of 6.

Set those against the busiest dockets, where most motions in the country are actually decided: S.D. Florida (187 decisions, 12.8%), M.D. Florida (147, 9.5%), E.D. Virginia (135, 12.6%), W.D. Missouri (114, 3.5%), and E.D. North Carolina (103, 5.8%). All five granted below the national rate. The places where the most people file are not the places most likely to say yes.

Put the extremes side by side and the spread is stark: D. Maryland granted 30 of 95, and the Southern District of Iowa granted 0 of 81. Two federal districts, the same statute, the same guideline, the same year.

Which Court Decides: the Sentencing Court, Not the Prison’s District

This is the single most misunderstood point about the data, and getting it wrong costs families months.

A compassionate release motion is filed in the federal district court that imposed the sentence. It is a motion to modify that court’s own judgment under 18 U.S.C. § 3582(c)(1)(A) — not a new case, and not a challenge to the conditions of confinement. The Bureau of Prisons’ own regulation routes its internal review to the U.S. Attorney “in the district in which the inmate was sentenced” and any resulting motion to “the sentencing court” (28 C.F.R. § 571.62(a)(3)).

So the two halves of the process sit in different places. The request to the warden goes to the facility where the person is confined right now, wherever the Bureau has moved them (28 C.F.R. § 571.61). The motion goes back to the courthouse where the case was prosecuted. Someone held at a federal prison in West Virginia but sentenced in Minnesota asks the West Virginia warden and files in Minnesota. A transfer does not move the motion, and it does not move you into a different set of numbers.

Which means a family cannot shop for a better district. The column of this page that applies to you was fixed years ago, at indictment. What you can control is the record you put in front of that court — which is exactly why the variance is worth understanding and not worth agonizing over. The procedural sequence, including the 30-day rule and what the written request has to contain, is covered on our page about the compassionate release request to warden.

Why Grant Rates Vary So Much Between Courts — and What the Variance Does Not Prove

A sixfold spread on identical statutory text is real, and it is worth saying plainly what it does and does not establish.

What it does establish. Different circuits read USSG § 1B1.13 differently, and circuit precedent binds every district court beneath it. The Supreme Court documented that split itself in 2026. In Rutherford v. United States, 146 S. Ct. 1320 (2026), the Court wrote that “[m]ost Courts of Appeals held that a nonretroactive amendment to a mandatory minimum cannot serve as an ‘extraordinary and compelling’ reason for a reduced sentence,” while “[a] minority … concluded that the disparity produced by such an amendment can justify a reduced sentence — at least when considered in combination with other factors” (slip op. at 5). Footnotes 1 and 2 name the circuits on each side: the majority view came from the Third, Fifth, Sixth, Seventh, Eighth, and D.C. Circuits; the minority view from the First, Fourth, Ninth, and Tenth.

Now set that against the FY2025 grant rates. Every one of the four minority-view circuits granted at or above the national rate — Ninth 31.0%, First 25.0%, Tenth 20.8%, Fourth 14.2%, against a national 14.1%. Five of the six majority-view circuits granted below it — Third 12.6%, Fifth 9.9%, Sixth 8.9%, Seventh 8.1%, Eighth 4.9% — with the D.C. Circuit the lone exception at 25.0%, on 20 decisions.

That alignment is not proof of causation, and the Rutherford question was only one of the issues these motions raise. But it is the clearest available account of why the same statute produces such different numbers, and it comes from the Supreme Court’s own description of the law rather than from anyone’s inference.

What it does not establish. Four honest caveats, and any page that gives you the spread without them is selling you something.

1. The Commission codes outcomes, not case quality. Its tables record what the court decided and the reasons the court gave. Nothing in the data measures how strong the underlying motions were, whether they were counseled, or how well documented they were. A district with a low rate may be seeing weaker filings, not applying a harsher standard.
2. Case mixes differ. Districts differ in offense composition, sentence lengths, the age of their prison populations, and how many people they sentenced during the eras that produced very long terms. A district whose docket is heavy with recent, short sentences has fewer people who can meet any version of the standard.
3. Filing rates differ, and the Commission does not publish them. These tables count motions decided, not motions filed. A district where few people file will show a small denominator and a volatile rate for reasons that have nothing to do with the judges.
4. Small numbers lie. By our count of the Commission’s district table, 48 of the 94 federal districts decided fewer than twenty of these motions in FY2025. At that scale a rate describes a handful of rulings, not a property of the court.

The honest summary: venue is a real input, and it is not the input you control. It should change your expectations and the depth of your record. It should not change whether a documented, qualifying case gets filed.

Applied Insight — Elizabeth Franklin-Best, Esq., Principal Attorney: The base rate is the wrong thing to argue with and the right thing to plan around. In a district granting one motion in thirty, assume the judge will reach the 3553(a) factors and assume they will decide that step against you unless the filing gives them a reason not to. That means briefing the sentencing factors affirmatively and on the current record — what the disciplinary history actually shows, what has happened in the years since sentencing, what supervision conditions could substitute for continued custody — rather than treating them as a closing paragraph after the medical argument. Nationally, courts cited § 3553(a) 1,260 times as a reason to deny in FY2025, more than the next two reasons combined. A motion that proves the illness and stops has done half the work in a district where half is not enough.

Who Gets Relief: What the Data Says About the People Whose Motions Were Granted

The Commission publishes a demographic and offense profile of the people whose motions succeeded. Five findings are worth knowing before you file.

Almost all of it is self-help. Of the 394 FY2025 grants, 369 (93.7%) were defendant-filed; the BOP Director filed 16 (4.1%), 9 were joint (2.3%), and the government filed none (FY2025 report, Table 5). In the first half of FY2026 it was 123 of 125 (98.4%). Across FY2020–22 the Director filed 45 of 4,491 granted motions (1.0%). Waiting for the Bureau of Prisons to act on its own is not a strategy — the statutory route Congress opened in 2018 is doing essentially all of the work.

The people who succeed are older, but not by much. Average age at original sentencing was 38 for both granted and denied movants. Average age at the decision on the motion was 51 for those granted and 47 for those denied (Table 6). A four-year difference. Age helps at the margin; it is not the dividing line, and the specific age-based ground under USSG § 1B1.13(b)(2) has its own requirements, covered on our page for compassionate release for elderly inmates.

Half were serving twenty years or more. Of the 394 grants, 50.0% had an original sentence of 20 years or longer, another 25.8% were serving 10 to under 20 years, 9.8% five to under ten, 10.8% two to under five, 3.1% one to under two, and 0.5% under a year (Figure 2; sentence-length information was missing in six cases). Read that figure for exactly what it says: it is titled “Original Sentence Length for Individuals Receiving Compassionate Release,” and it is the length of the sentence they were already serving, not the size of any reduction. The Commission publishes no median or average reduction anywhere in this series, so anyone quoting you an average sentence cut is not quoting the Commission.

Criminal history is not the filter people assume. Criminal history category VI accounted for 35.7% of grants and 36.3% of denials; category I accounted for 27.5% of grants and 23.1% of denials (Table 7; these rows rest on the table’s base of 2,785 cases). Career offenders under § 4B1.1 were 20.7% of grants and 18.4% of denials. On the guideline-application subset of 2,664 cases with complete information, a weapon was involved in 48.8% of grants and 44.4% of denials, and a firearms mandatory minimum applied in 37.6% of grants and 27.0% of denials. Serious criminal history and firearms are not disqualifiers in the data; they show up at nearly the same rate on both sides of the line.

By offense, the 390 grants with offense information were 46.4% drug trafficking (181), 16.2% robbery (63), 13.3% firearms (52), 6.7% murder (26), and 5.4% fraud, theft, or embezzlement (21) (Table 8). Child pornography accounted for 0 of 390 grants and 82 of 2,386 denials (Table 9). Section 3582(c)(1)(A) contains no offense-based exclusion — but the data shows some offenses are effectively out of reach anyway, which is covered on our page about who is not eligible for compassionate release.

What Courts Credit When They Grant, and What They Deny On

The Commission codes the reasons courts give, in Tables 10 and 11. Courts may give more than one reason per case, so these are shares of reasons, not shares of cases. This is the single most-misreported statistic in this whole dataset, and it is worth stating twice: 1,260 § 3553(a) citations does not mean 1,260 motions were denied on § 3553(a) grounds.

The reasons courts gave for granting

In FY2025, 394 cases were granted and 631 reasons were recorded.

Reason given for granting	Citations	Share of the 631 reasons
Rehabilitation	95	15.1%
Unusually long sentence and change in law, USSG § 1B1.13(b)(6)	80	12.7%
Serious physical or medical condition, § 1B1.13(b)(1)(B)	71	11.3%
Missing / not coded	53	8.4%
Multiple 18 U.S.C. § 924(c) penalties	45	7.1%
Other mandatory minimum penalties / long sentence	31	4.9%
Terminal illness, § 1B1.13(b)(1)(A)	28	4.4%
Family circumstances — care for a parent, § 1B1.13(b)(3)(C)	23	3.6%
Young age at time of offense	19	3.0%
Family circumstances — care for a minor or disabled child, § 1B1.13(b)(3)(A)	18	2.9%
21 U.S.C. § 851 enhanced drug penalties	18	2.9%
Age 65 with deteriorating health and time served, § 1B1.13(b)(2)	16	2.5%
BOP failure to provide treatment, § 1B1.13(b)(1)(C)	12	1.9%
Abuse by a corrections officer or BOP contractor, § 1B1.13(b)(4)	5	0.8%
COVID-19 / pandemic, § 1B1.13(b)(1)(D)	3	0.5%
Other	41	6.5%

Source: USSC FY2025 Data Report, Table 10. Selected rows; the table lists 27 categories.

Two things to take from it. First, the Commission’s own footnote: “In all cases where the court gave rehabilitation as a reason for the granted motion, the court also gave one or more other reasons.” Rehabilitation is the most-cited reason and it never stood alone — consistent with 28 U.S.C. § 994(t), which forbids treating rehabilitation by itself as extraordinary and compelling. Second, COVID-19 is functionally gone: 3 citations in FY2025, against 1,304 in FY2020, when it was 61.6% of all granting reasons.

The reasons courts gave for denying

In FY2025, 2,401 cases were denied and 5,067 reasons were recorded.

Reason given for denying	Citations	Share of the 5,067 reasons
18 U.S.C. § 3553(a) factors	1,260	24.9%
Rehabilitation insufficient	524	10.3%
No extraordinary and compelling reason provided	512	10.1%
Failure to exhaust administrative remedies	447	8.8%
Insufficient proof of a serious physical or medical condition	352	6.9%
Defendant able to provide self-care in prison	260	5.1%
Not at risk from COVID / pandemic	191	3.8%
Protection of the public	143	2.8%
Other care available for a minor child	142	2.8%
Does not meet the age or time-served requirement under 18 U.S.C. § 3559	120	2.4%
Insufficient proof of deteriorating physical or mental health	66	1.3%
Post-sentencing or post-conviction conduct	55	1.1%
Other	882	17.4%

Source: USSC FY2025 Data Report, Table 11. Selected rows.

The first half of FY2026 looks the same in shape: across 769 denied cases and 1,752 reasons, § 3553(a) drew 403 citations (23.0%), no extraordinary and compelling reason 235 (13.4%), rehabilitation insufficient 207 (11.8%), failure to exhaust 136 (7.8%), insufficient medical proof 118 (6.7%), able to self-care 107 (6.1%), and protection of the public 47 (2.7%) (FY2026 Q1–Q2 report, Table 11).

Three of these are avoidable before the motion is written. Failure to exhaust — 447 citations — is a calendar problem, not a merits problem. Insufficient proof of a serious medical condition (352 citations) and able to provide self-care in prison (260) are records problems; USSG § 1B1.13(b)(1)(B) asks whether a condition “substantially diminishes the ability of the defendant to provide self-care within the environment of a correctional facility,” which is a functional question the Bureau’s own files usually answer. That evidentiary fight is the subject of our page on compassionate release medical condition. Other care available for a minor child — 142 citations — is the standard failure mode in family-based motions, covered under caregiver compassionate release.

A caveat the Commission’s own tables force. “Missing” accounts for 8.4% of granting reasons, and “Other” for 17.4% of denial reasons. The reason data is a strong signal about what courts talk about; it is not a complete census of why they rule.

Applied Insight — Christopher Zoukis, JD, MBA, Managing Director: Look at the two denial lines that sit next to each other — insufficient proof of a serious medical condition, 352, and able to provide self-care in prison, 260. Those are not rulings about how frightening a diagnosis is. They are rulings about function, and the evidence that answers them is almost always already inside the Bureau of Prisons’ own systems before anyone files anything: the Care Level assignment, the Activities of Daily Living assessment, the chronic care clinic notes, the medical duty status, the assistive-device authorizations, the housing accommodations the institution already made. Request the complete health record, not the summary a family can get over the phone, and build the timeline before deciding which subsection of the guideline the case fits. When the government’s own paperwork says a person needs help to function, that carries further with a judge than an outside letter saying the same thing. The legal work — which subsection, how it is briefed, and the filing itself — is handled by Elizabeth Franklin-Best and the firm’s attorneys.

Two Numbers About to Fall Out of the Grant Column

Every figure elsewhere on this page predates the Supreme Court’s decision in Rutherford v. United States: it comes from reports through FY2026 Q2, which covers October 1, 2025 through March 31, 2026 and reflects motions analyzed through April 27, 2026. Rutherford and its companion Fernandez v. United States were decided May 28, 2026 — after that window closed. The Commission’s September 2026 report, covering decisions through June 30, 2026, is the first to include any post-Rutherford rulings — about one month of them. For FY2026 through June it shows 1,432 motions decided and 195 granted (13.6%), and June 2026 alone shows 18 grants among 179 motions decided (10.1%) (Tables 1–2). One month is not yet a measure of what the decisions did to grant rates.

That matters because two of the reasons courts cited most often when granting motions describe a theory the Court has now rejected.

Granting reason	FY2024	FY2025	FY2026 Q1–Q2
Unusually long sentence + change in law, § 1B1.13(b)(6)	98 citations (13.0% of 753 reasons)	80 (12.7% of 631)	17 (8.5% of 199)
Multiple 18 U.S.C. § 924(c) penalties	51 (6.8% of 753)	45 (7.1% of 631)	11 (5.5% of 199)

Source: Table 10 of the FY2024, FY2025, and FY2026 Q1–Q2 reports.

In FY2025 those two rows together were 125 of the 631 recorded granting reasons — our addition of two published rows, not a figure the Commission prints. Both describe grants premised on the idea that a sentence left long by a change Congress declined to make retroactive can be an extraordinary and compelling reason.

Rutherford held it cannot. The nonretroactive change, “considered by itself or in combination with other factors,” cannot make a prisoner eligible, and “[t]o the extent that it counsels otherwise, the Commission’s policy statement is invalid” (slip opinion at 16). Section 1B1.13(b)(6) is still printed in the current Guidelines Manual and the Commission has not amended it, so a reader who looks it up will find a provision the Supreme Court has held invalid in significant part, with nothing on the page to say so.

The practical consequence for anyone using this page to estimate odds: any figure built on FY2024 or FY2025 data overstates what is available today. A meaningful share of the grants counted in those years rested on a theory that no longer works. The share was already falling before the decision — 13.0% to 12.7% to 8.5% — though the Commission publishes no analysis of why, and the drop through March 2026 cannot be attributed to a ruling issued two months later.

What the decision did not do matters too. The Court expressly reserved the broader question: “We need not decide whether there are reasons beyond personal circumstances that could qualify as ‘extraordinary and compelling.'” (Slip op. at 10–11.) And footnote 3 says the 2023 amendment’s victim-of-abuse provision “is not at issue in today’s cases,” leaving USSG § 1B1.13(b)(4) untouched — which matters to people harmed in custody, covered on our page for victims of sexual abuse lawyer. The full treatment of both 2026 decisions is on our rutherford v united states page, and the current state of the unusually long sentence provision has its own page.

How to Read These Numbers, and Where They Come From

Six rules for using the data without overreading it

Stated as rules because that is what they are.

1. Never quote a rate without its denominator. “The First Circuit grants 47.6%” is a sentence that will get somebody’s hopes up over 21 decisions. “10 of 21 in six months” is the same fact and cannot mislead anyone.
2. Never mix reporting periods. A FY2025 district figure and a FY2026 half-year circuit figure do not belong in the same comparison. Cite each to its own report.
3. Remember these are decisions, not filings. The Commission does not publish how many motions were filed, and a motion still pending on the last day of a reporting period is not in the tables at all. Timing questions are covered on how long does compassionate release take.
4. Reason percentages are shares of reasons. Courts gave 5,067 reasons across 2,401 denials in FY2025 — roughly two apiece.
5. Figure 2 is the original sentence, not the reduction. The Commission publishes nothing about how much time a grant takes off.
6. Treat every pre-May-2026 grant that rested on sentence length as historical. See the section above.

And one rule that is about judgment rather than arithmetic: the base rate is a description of a court, not a prediction about a case. A well-documented terminal-illness motion in the Southern District of Iowa is a better motion than a thin one in Maryland, and no percentage on this page changes that. What the numbers should change is the level of proof you plan for — the completeness of the medical record, the specificity of the compassionate release plan, the evidence answering the danger to the community finding, and how affirmatively the sentencing factors are briefed. Those are the variables you control, and in a low-granting district they are the only ones. If a motion has already been denied, the analysis of what to do next lives on compassionate release denial, and the standard for a renewed filing on can you file compassionate release twice.

Where the numbers come from, and when they update

Everything on this page comes from the U.S. Sentencing Commission’s Compassionate Release Data Reports, a public series the Commission maintains from the sentencing documents courts are required to send it under 28 U.S.C. § 994(w). As of September 2026 the series consists of six reports:

Report	Period covered	Published
FY2020–FY2022 combined	Oct. 1, 2019 – Sept. 30, 2022	December 2022
Fiscal Year 2023	Oct. 1, 2022 – Sept. 30, 2023	March 2024
Fiscal Year 2024	Oct. 1, 2023 – Sept. 30, 2024	March 2025
Fiscal Year 2025	Oct. 1, 2024 – Sept. 30, 2025	May 2026
FY2026 cumulative through Q2	Oct. 1, 2025 – Mar. 31, 2026	May 2026
FY2026 cumulative through Q3 — newest	Oct. 1, 2025 – June 30, 2026	September 2026

A federal fiscal year runs October 1 through September 30. The Commission publishes cumulative preliminary data during an ongoing fiscal year and a full report after it closes, which in practice means this page has two update points a year. The quarterly reports are expressly labeled preliminary.

Each report states its own data cutoff — the FY2025 and FY2026 Q2 reports were both limited to motions the Commission had analyzed by April 27, 2026. Late-arriving documentation is why earlier years are restated upward in later reports, and why a figure should always be cited to the report it came from rather than to “the Sentencing Commission” in the abstract.

What the Commission does not publish, and what therefore cannot honestly appear on any page: the number of motions filed; any median or average sentence reduction; single-year FY2020, FY2021, or FY2022 totals or rates; a breakdown of § 1B1.13(b)(5) as a discrete coded category; and any analysis of what Amendment 814 or the 2026 decisions did to grant rates. If you see one of those attributed to the Commission, it is somebody’s estimate.

Why work with Elizabeth Franklin-Best, P.C.

Elizabeth Franklin-Best is the founder and Principal Attorney of Elizabeth Franklin-Best, P.C., a boutique firm devoted to federal criminal appeals and post-conviction relief. Over more than 20 years she has handled more than 330 federal proceedings, over 100 of them appeals, with sixteen published decisions as counsel of record. She is admitted to the U.S. Supreme Court and twelve of the thirteen U.S. Courts of Appeals — every regional circuit (1st–11th) plus the D.C. Circuit — and has appeared pro hac vice in federal courts from Montana to Florida. She is recognized by Best Lawyers in America 2026 for Appellate Practice and by Chambers USA 2026 for Litigation: White-Collar Crime & Government Investigations, and was named South Carolina Public Defender of the Year in 2010. She is the author of Reversing Your Criminal Conviction: Develop Your Winning Strategy (2023) and serves as Board Chair of Justice 360.

Christopher Zoukis, JD, MBA, is the firm’s Managing Director and a nationally recognized federal prison expert. He is the author of the Federal Prison Handbook, Directory of Federal Prisons, Prison Education Guide, and College for Convicts, and serves as Chairman of the Board of the Human Rights Defense Center. He has been quoted as a federal prison expert by outlets including CNN, the Associated Press, and The Washington Post. He earned his J.D. at the University of California, Davis School of Law after serving roughly a decade in the federal Bureau of Prisons, where he became an award-winning incarcerated journalist — a combination of formal training and firsthand knowledge that few in federal defense can offer.

The firm runs a national federal practice on an intentionally small caseload. Compassionate release motions are filed in the district that imposed the sentence, wherever in the country that court sits — which is why we read the district’s record, and the assigned judge’s, before advising anyone about what a filing is worth. If you are still working out what this relief is and whether it fits, start with the compassionate release definition or the comparison of routes on compassionate release attorneys. If the question is what the judge will actually weigh, see what do judges consider when sentencing. If the real problem is that the conviction itself was wrong, the vehicle after Fernandez v. United States, 146 S. Ct. 1292 (2026), is a § 2255 motion, not this one — the line between the two is drawn on compassionate release vs 2255.

Talk to a federal compassionate release lawyer

A grant rate tells you what a court has done. It cannot tell you what your record supports. The useful first step is a look at the actual file — the judgment, the district, the assigned judge, the medical record, and what has changed since sentencing.

Call (843) 620-1100 or schedule a federal compassionate release case review with our federal compassionate release lawyers.

Frequently Asked Questions About Compassionate Release Grant Rates

Reviewed for legal accuracy by Elizabeth Franklin-Best, Esq., Principal Attorney·September 2026