by Richard Resch

Six years after Congress passed the First Step Act with rare bipartisan enthusiasm, the federal government’s own data offer a mixed picture of progress. The Bureau of Justice Statistics’ (“BJS”) March 2026 report, covering calendar year 2024, provides the most comprehensive look yet at who sits in federal prison, what programming they receive, and how the system measures their readiness to return to society. Some of the numbers reflect genuine advances. Others suggest that the 2018 law’s promise remains largely unfulfilled.

The data deserve close attention from anyone who is interested in federal criminal justice policy, because they reveal both what the government is willing to count and what those counts actually mean for the 154,093 individuals held in Bureau of Prisons (“BOP”) custody at the end of 2024.

A Shrinking but Still
Massive Population

The federal prison population dropped by roughly 1% between yearend 2023 and yearend 2024, declining from 155,972 to 154,093. That modest decrease continues a trend that began after the system peaked at 158,637 in 2022. To put the number in perspective, the federal system still holds more people than the entire prison populations of most Western democracies.

The demographic contours of this population remain largely stable. According to the BJS report, 53.1% of federal prisoners have never been married, a figure that has held constant across all five years of reported data. An estimated 21.4% are married. These estimates, it should be noted, are not drawn from current surveys but extrapolated from the BJS’s 2016 Survey of Prison Inmates, applied to current population totals and rounded to the nearest hundred. The methodology means the marital status data are approximations, a limitation worth keeping in mind.

Non-U.S. citizens accounted for 14.2% of the federal prison population at yearend 2024, down from 15.9% in 2020. The number of non-citizens in BOP custody fell from 24,009 to 21,948 over that period. English-language learners made up about 10% of the population, with 15,501 individuals for whom English was a second language.

One statistic stands out for its real-world impact. At yearend 2024, 30.5% of federal prisoners had a minor child. That figure has plummeted from 51.8% in 2020, but the BOP defines “minor child” as a person age 20 or younger listed as a dependent, so some of this decline reflects children aging out of the definition rather than any change in family circumstances. Even at 30.5%, the number means that nearly 47,000 people behind federal bars are parents of dependents, a reminder that incarceration’s consequences extend well beyond prison walls.

A Sharp Rise in
Addiction Treatment

If any single data point in the 2026 report warrants attention, it is the dramatic expansion of medication-assisted treatment (“MAT”) for substance use disorders. In 2024, a total of 12,479 people received FDA-approved MAT while in BOP custody, representing a 112% increase from the 5,898 who received such treatment in 2023. The trajectory is remarkable. In 2020, the number was just 418.

The growth in MAT reflects a broader shift in how the federal system addresses addiction. The number of people who were already receiving MAT before entering federal custody also rose sharply, from 195 in 2020 to 2,935 in 2024, suggesting that courts and pretrial services are increasingly connecting defendants with treatment that the BOP is then continuing rather than disrupting.

These are encouraging numbers. For decades, correctional systems treated addiction primarily through abstinence-based programming, despite overwhelming clinical evidence that medications like buprenorphine, methadone, and naltrexone dramatically reduce overdose deaths and recidivism. The federal system’s belated embrace of MAT represents one of the First Step Act era’s more tangible achievements.

The traditional drug treatment programs tell a more complicated story. Volunteers for the Non-Residential Drug Abuse Program grew from 19,514 in 2020 to 44,777 in 2024, but actual participation in the program dropped slightly from 21,755 in 2023 to 20,556 in 2024. The gap between volunteers and participants is revealing. More than twice as many people wanted to participate as actually did. The Residential Drug Abuse Program showed a similar pattern, with 12,111 volunteers but only 11,837 participants. These gaps suggest capacity constraints that limit access to programming the law was designed to expand.

PATTERN and the Racial Divide
in Risk Assessment

The First Step Act’s signature innovation was the Prisoner Assessment Tool Targeting Estimated Risk and Needs, known as PATTERN, which classifies federal prisoners by their estimated recidivism risk and is supposed to guide their access to earned time credits and programming. By yearend 2024, a total of 141,209 people had been assessed. Another 13,430 had not received a PATTERN score, typically because they were in pretrial detention or had not been in custody long enough.

The topline numbers look reasonable enough: 55% of assessed prisoners were classified as minimum or low risk, 19% as medium risk, and 26% as high risk. The gender disparity is striking, with 83% of women classified as minimum or low risk compared to 53% of men.

The racial disparities in PATTERN scores, however, merit scrutiny. According to the BJS data, 59% of Black federal prisoners and 58% of American Indian or Alaska Native prisoners were classified as medium or high risk for recidivism. For White prisoners, that figure was 35%. For Asian, Native Hawaiian, or Other Pacific Islander prisoners, it was 23%.

These disparities raise questions that the data alone cannot answer. PATTERN scores are calculated using factors including criminal history, education level, and disciplinary records. If those inputs themselves reflect systemic inequities in policing, prosecution, and access to education, then the tool may be encoding existing disparities into its risk predictions rather than measuring something intrinsic about the individuals being assessed. The fact that PATTERN’s own validation studies show different recidivism base rates by risk level does not resolve the concern, because recidivism itself is measured through rearrest and reimprisonment, outcomes influenced by the same systemic factors.

Age provides an instructive comparison. Among prisoners ages 55 to 64, fully 83% were classified as minimum or low risk. Among those 65 and older, the figure was 93%. This aligns with decades of criminological research showing that the risk of reoffending declines sharply with age. If the PATTERN tool captures age-related desistance accurately, why do thousands of elderly prisoners classified as presenting minimal risk remain incarcerated? According to the data, 5,039 people age 65 or older had been assessed, and the vast majority posed little measurable risk. Their continued confinement raises questions about whether the First Step Act’s risk assessment tools are actually being used to drive release decisions or merely to generate data.

How Prison Discipline
Shapes Time Credits

The data on prohibited acts illuminate how the BOP’s internal disciplinary system operates, and how it intersects with the First Step Act’s earned time credits. In 2024, federal prisoners committed 103,268 prohibited acts that resulted in reductions in rewards, incentives, or time credits. A total of 57,006 individuals were responsible for those acts, meaning that a significant number of people were cited multiple times.

Medium security prisons accounted for a disproportionate share of disciplinary actions. Although they represent one security tier among five, medium security facilities generated 45.9% of all prohibited acts and 47.9% of those classified at the greatest severity level. High security prisons, which house the most restricted populations, accounted for 18% of total acts but 28.8% of high-severity offenses.

The racial breakdown of disciplinary citations raises its own concerns. White prisoners accounted for 51.6% of those cited, while Black prisoners accounted for 43.5%. Without knowing the racial composition of each facility’s population and the specific circumstances underlying each citation, it is impossible to determine whether these rates reflect differences in behavior, differences in enforcement, or some combination. But the data, read alongside the PATTERN racial disparities, suggest that the disciplinary system warrants independent examination for potential bias.

The facilities with the highest total prohibited acts in 2024 were Schuylkill FCI in Pennsylvania (1,904), Brooklyn MDC in New York (1,817), and Canaan USP in Pennsylvania (1,801). Edgefield FCI in South Carolina recorded the most acts at the greatest severity level, with 849. These facility-level variations are enormous and likely reflect differences in management, staffing, population characteristics, and institutional culture. Whether the BOP is examining these outliers and addressing their causes is a question the data do not answer.

Segregated Housing and
Solitary Confinement

The First Step Act specifically requires the BOP to report the number of prisoners placed in solitary confinement. The 2024 figures are sobering. The average daily population in special housing units (“SHU”), where individuals are separated from the general prison population, was 11,827. That figure has remained stubbornly high across the reporting period. After dropping from 10,236 in 2020 to 9,261 in 2021, the average daily SHU population climbed to 11,974 by 2023 before dipping slightly in 2024.

The average daily population in administrative maximum housing, the most restrictive form of segregated confinement in the federal system, was 337 in 2024. The BOP operates one such facility, the United States Penitentiary-Administrative Maximum in Florence, Colorado.

The BOP’s only special management unit, located at Thomson USP, was shuttered in February 2023 after the agency identified what the BJS report describes as “significant concerns” at the facility.

Nearly 12,000 people held in segregated housing on any given day is a figure that warrants sustained attention, given the extensive body of research documenting the psychological harm that prolonged isolation inflicts on incarcerated individuals.

Healthcare Vacancies and Accreditation Losses

Buried in a table of facility characteristics are two data points that should alarm anyone concerned about conditions of confinement. The average vacancy rate for medical and healthcare staff positions at BOP facilities reached 22% in 2024, up from 9% in 2020 and 21% in both 2022 and 2023. One in five healthcare positions sitting empty in a system holding more than 154,000 people is not a minor administrative inconvenience. It is a crisis of access to care.

The report notes that zero facilities operated without at least one clinical nurse, paramedic, or licensed physician on site. But having one clinician present does not mean a facility is adequately staffed to meet constitutional standards for medical care. A 22% vacancy rate means longer wait times, delayed diagnoses, and reduced access to mental health services for a population with significant healthcare needs.

The accreditation picture deteriorated as well. The BOP’s contract with the American Correctional Association (“ACA”) expired on March 31, 2024, without renewal. At the point of contract expiration, all BOP facilities were ACA accredited. But as individual facility accreditations began expiring in sequence, the number of ACA-accredited facilities fell to 99 by yearend 2024. ACA accreditation has long been criticized as insufficiently rigorous, but its loss removes even that thin layer of external oversight. As of yearend 2024, ACA accreditation had expired for 22 BOP facilities.

Pregnant Women in
Federal Custody

The data on pregnant women in federal custody, while involving small numbers, highlight the conditions facing some of the system’s most vulnerable people. In 2024, 138 pregnant women in BOP facilities had known pregnancy outcomes by December 31. Fifty percent of those pregnancies ended in live birth. Forty-one percent of the women were released before giving birth, meaning their outcomes were unknown to the system.

Restraints were used on these women four times during 2024, with all four incidents occurring during postpartum recovery. The use of any restraints on postpartum women raises concerns about compliance with the First Step Act’s own restrictions on such practices, which were enacted in response to well-documented evidence that shackling pregnant and postpartum women poses serious medical risks.

Programming: Breadth
Without Depth

On paper, the BOP’s programming portfolio is extensive. The system identified 51 evidence-based recidivism reduction programs and 71 productive activities available to prisoners in 2024. These ranged from anger management and cognitive behavioral therapy to literacy programs, vocational training, and financial education. Many of the core programs were available at all 121 facilities.

The BOP maintained 4,114 external partnerships to provide recidivism reduction programming, with 60% of those partnerships involving faith-based organizations. The heavy reliance on faith-based groups raises questions about whether the programming available to federal prisoners is sufficiently varied to meet diverse needs and whether secular alternatives are equally accessible.

The number of registered volunteers stood at 4,519, with the vast majority (3,720) classified as Level II volunteers who serve five or more days per year. Some facilities had robust volunteer programs. For instance, Fort Dix FCI had 193 volunteers. Others had almost none. The variation suggests that access to community-based programming depends heavily on geography and institutional initiative.

The Limits of the Data

The First Step Act’s data collection mandate was itself a reform. Before 2018, much of this information was either not collected or not publicly reported. The annual BJS reports represent genuine transparency, and the trends they capture are valuable for researchers, advocates, and policymakers.

But transparency about inputs is not the same as accountability for outcomes. The report does not track how many people have actually earned time credits under the First Step Act, how many have been released early as a result, or what happens to them after release. It does not examine whether PATTERN’s risk classifications are being used to expand or restrict access to programming. It does not evaluate whether the BOP’s 51 approved programs are adequately funded, properly staffed, or producing measurable results.

The data show that 26% of federal prisoners are classified as high risk for recidivism and that 40% of those under age 25 fall into that category. They do not show whether those individuals are receiving the intensive programming that research indicates can change their trajectories, or whether they are simply warehoused with a risk label attached.

Six years in, the First Step Act’s data collection requirements were working. The harder question is whether anything else about the law is working as intended. The numbers in this report provide the raw material for that assessment. The assessment itself remains the work of legislators, courts, and the public.  

Source: Bureau of Justice Statistics, “Federal Prisoner Statistics Collected Under the First Step Act, 2025,” NCJ 310701, March 2026.

Subscribe today

Already a subscriber? Login