First Step Act release estimator

Good conduct time is 54 days per year of the sentence imposed and reduces the sentence itself. FSA credits do not shorten the sentence. Instead they become applicable once the earned balance equals the days remaining, and up to 365 of them may go toward early transfer to supervised release.

* Programming means being enrolled in, or on a waiting list for, classes identified as a need by the needs assessment. Waitlist time counts as earning status for up to two assessment periods, so long as nothing has been refused or declined; active participation in any one program removes that limit. The needs assessment (SPARC-13) can be completed on a computer the first day, and getting it done early matters: until it is finished the person counts as opted out and earns nothing, regardless of eligibility. Declining a program after sitting on its waitlist rescinds the credits earned during that wait.

Only official detention counts as jail credit. Section 3585(b) covers time in custody, meaning jail, a detention center, or a transfer facility. Restrictive release conditions do not qualify regardless of severity: home detention, electronic monitoring, curfews, halfway house placement as a condition of bond, inpatient treatment on release. Reno v. Koray settled this. Koray was confined to a halfway house 24 hours a day for 150 days under an order barring him from leaving for any reason unless accompanied by a federal agent, counted five times daily, subject to random breath and urine tests. The Supreme Court held that was still "release," not detention, and he got nothing for it. Time already credited against another sentence also cannot be counted twice.

† Time before arrival at the designated facility earns nothing. The program statement is explicit that credits cannot be earned in U.S. Marshals custody prior to arriving at the designated facility, regardless of where the person is housed. Someone remanded at sentencing may wait weeks or months in a county jail or transfer center before designation, and none of it counts. That time is credited against the sentence under § 3585(b), but the FSA credits it would otherwise have produced are gone and cannot be recovered later.

This has been litigated without much success. A magistrate judge in the District of New Hampshire ordered transport-period credits awarded in a 2023 case, and the prisoner was released the next day, but that ruling did not become the general rule for anyone else. On the related question of transfers between institutions, the Fourth Circuit held in January 2026 that a prisoner housed three days in a transfer center's SHU while moving between prisons was not entitled to the credit they would otherwise have earned, over a dissent arguing the majority relied on a theory the government never raised below. Treat this as unresolved but currently going the BOP's way.

Credits post in whole 30-day blocks. Nothing appears on the record until a full 30 days in earning status completes, and partial credit is not awarded, so leftover days carry into the next monthly cycle. Non-earning status (disciplinary segregation, writ, outside placement for a full day, opting out, refusing the SPARC-13 or a required program) pauses accrual entirely.

Credits are not subtracted in a lump at the end. Under 28 CFR § 523.44(b)(1) they become applicable once the earned balance equals the remainder of the imposed term. The balance rises while the remaining term falls, and release comes at the crossing point.

Surplus past 365 days is real but uncommon. Policy directs that prerelease placement in an RRC or on home confinement be based on FTCs other than those already applied to supervised release, so a surplus does have somewhere to go. In practice most people never build one: of 17,894 released in 2024 after applying credits, 11,557 applied them toward supervised release only and averaged 5 months of credits total. The 6,337 who got credits toward both averaged 24 months of credits on longer sentences. A surplus is a long-sentence outcome, not a default.

Credits accrue regardless of PATTERN score, because a medium or high score stops them from being applied, not earned. But medium and high receive no projected FSA release date at all, so there is no date to plan against while the balance builds.

The BOP does not assume a score will improve. It works from where the score is now, so a projection built on eventually reaching minimum/low is a best case rather than a date to plan around. Scores also move upward on review, which can strand a balance that was previously on track to be applied.

RDAP is a separate reduction under 18 U.S.C. § 3621(e), available to non-violent offenders who complete the program including its community-based component. The § 3621(e) benefit applies first, then FTCs. If there is not enough time left to take both, the FTC days applied are cut back to preserve at minimum the 120-day community placement RDAP requires. The full 12 months is a ceiling, not an entitlement.

Do not plan around a year of halfway house time. The Second Chance Act permits the BOP to grant up to 12 months of prerelease custody, but 18 U.S.C. § 3624(c) has never required it. In March 2025 the BOP tried to cap SCA placement at 60 days, citing budget pressure; it rescinded that a week later, and a June 2025 directive from the Director said FSA and SCA authorities are cumulative and that bed availability should not block home confinement. In practice, placements have continued to run far short of a year.

The binding constraint is upstream of any of this. The BOP contracts for a fixed number of RRC beds and has not expanded that capacity to match what the First Step Act made people eligible for. It has cancelled solicitations for additional beds over budget and staffing concerns. The program statement says as much outright: prerelease placement depends on release residence, program requirements, and available contract bed space and funding. Statutory eligibility does not create a bed. A directive saying capacity "shall not be a barrier" does not itself build capacity.

Real dates come from the BOP's own computation, not from this page.