The Sentencing Reform Act of 1984 is one of those laws that made sense on paper at the time it passed. Federal parole had genuine problems: inconsistent outcomes, different boards making wildly different calls on identical cases, disparity that tracked race and geography more than it tracked risk or rehabilitation. Congress looked at that mess and concluded the solution was to eliminate discretion from the back end of the system altogether. No more parole board. Offenders would serve the sentence the judge imposed, minus a capped amount of good time credit, and that would be the end of it.

What nobody adequately modeled was what this would cost, who it would hold, and what the legal exposure would look like when the government became the indefinite custodian of large numbers of people who, under any reasonable analysis, could be supervised in the community at a fraction of the cost and with no meaningful increase in public risk.

The record on that question is now four decades long. It is not ambiguous.

Parole, Probation, and the Sentencing Reform Act of 1984

Before November 1, 1987, the federal system operated under an indeterminate sentencing model. Judges handed down maximum sentences; the U.S. Parole Commission determined when a prisoner had served enough time and could be released under supervision. Most federal prisoners served roughly 58 percent of their maximum sentence before being paroled. The system was individualized, responsive to rehabilitation, and admittedly inconsistent.

The Sentencing Reform Act abolished parole for all federal offenses committed on or after November 1, 1987. In its place, Congress created the U.S. Sentencing Guidelines, a structured grid tying offense level to criminal history to produce recommended sentencing ranges. The guidelines were mandatory until the Supreme Court’s decision in United States v. Booker (2005) made them advisory, but their gravitational pull on federal sentencing remains substantial.

The secondary effect, which Congress understood but appears to have not fully weighted, was the near-elimination of probation as a standalone federal sentence. Before the guidelines, approximately half of all convicted federal offenders received probation. The guidelines made probation nearly unavailable for most offense categories, funneling convictions into prison terms that had to be served at 85 percent of their face value.

The result was structural: the federal prison population, which sat at roughly 24,000 in 1980, had grown to over 218,000 by the mid-2010s before reforms began to modestly reverse the trend. As of early 2025, the Bureau of Prisons holds approximately 155,000 people in federal facilities. Drug offenses alone account for approximately 45 percent of the federal inmate population, according to the Sentencing Project’s analysis of BOP data. Many of those are serving long mandatory minimum sentences with limited access to early release mechanisms.

What Abolished Parole Actually Means Inside a Sentence

The language people use for this is “truth in sentencing,” and it has a certain intuitive appeal. The judge says five years, the offender serves five years, the public knows what it got. But the practical mechanics are worth spelling out, because they explain why the policy costs what it costs and why modest reforms like the First Step Act of 2018 have moved the needle less than their supporters hoped.

Under the current system, a federal prisoner sentenced to ten years will serve roughly eight years and six months at minimum, assuming no disciplinary infractions that cost them good time credits. They have no mechanism to accelerate that release based on demonstrated rehabilitation, strong institutional conduct, completed programming, or changed circumstances. The Parole Commission, which performed those individualized assessments, was abolished. What replaced it was a formula.

The First Step Act of 2018 created an earned time credit system layered on top of good time credits. Eligible prisoners who complete evidence-based recidivism reduction programs can earn additional credits toward transfer to prerelease custody, meaning halfway houses or home confinement. This was a meaningful reform. But the operative word is “eligible.” Approximately 59,000 federal prisoners are categorically ineligible for earned time credits based on the nature of their offense. The BOP’s risk assessment tool, called PATTERN, has been repeatedly criticized for overestimating recidivism risk and producing racially disparate results. And even for eligible prisoners, the Council on Criminal Justice found in December 2024 that individuals released under the First Step Act in 2023 served an average of 82.1 percent of their imposed prison terms, compared to 89.7 percent before the law. Seven points. That is the scope of the reform to date.

The Fiscal Arithmetic of Abolished Parole

The Federal Register published the Bureau of Prisons’ official cost of incarceration fee for fiscal year 2024 in December 2025. The average annual cost of housing a federal inmate in a Bureau or non-Bureau facility is $47,162, or $129.21 per day. That is the direct custody cost. It does not include the cost of prosecution, the cost of pretrial detention, the cost of supervision after release, or the downstream costs that accumulate when a person spends extended time incarcerated and returns to the community with no stable employment, housing, or income.

The aggregate number is staggering. The Bureau of Prisons budget consumed over 25 percent of the entire Department of Justice budget by 2013, and projections from the Urban Institute at that time estimated it was tracking toward 30 percent by 2020. Funding an expanding federal prison population crowds out every other DOJ priority. Federal investigators. Prosecutors. Support for state and local governments. The entire ecosystem of federal law enforcement competing for budget space with the cost of housing people who, in many cases, committed non-violent offenses and could be supervised in the community at the probation rate.

That probation rate, documented by the Urban Institute: approximately $3,433 per offender per year. Community corrections, halfway houses, home confinement: approximately $43,703 per year. Federal prison: $47,162. The math is not complicated. The public is paying fourteen times more to house someone in a federal facility than to supervise them in the community, for a population that is predominantly non-violent, predominantly drug-offense, and serving sentences that no individualized review mechanism can shorten based on demonstrated risk reduction.

The Prison Policy Initiative estimates that when all costs are aggregated — policing, prosecution, courts, incarceration, supervision — the total annual cost of the American incarceration system approaches $182 billion. The Bureau of Justice Statistics puts the narrower figure at approximately $81 billion for corrections alone. Even using the conservative figure, the federal share of a policy that concentrates non-violent offenders in expensive facilities for the maximum possible time reflects a choice about resource allocation that the record does not support.

The Non-Violent Majority the Policy Was Not Designed For

One of the rhetorical foundations of the Sentencing Reform Act was the dangerous offender who was being let out early by a soft parole board. The public debate in the early 1980s was saturated with that image. The legislative response was calibrated to prevent early release of violent criminals by eliminating early release entirely.

The policy that emerged applied universally. Violent and non-violent offenders alike were subject to the same 85 percent floor, the same elimination of individualized review, the same mandatory minimum sentences that the guidelines imported and amplified. The drug sentencing provisions of the Anti-Drug Abuse Act of 1986, passed two years after the SRA, compounded this by imposing multi-year mandatory minimums for drug trafficking offenses that in many cases required no violence, no victim in the traditional sense, and no prior criminal history.

The result is the current federal prison population: roughly 45 percent drug offenses, the majority of those at the trafficking level but not involving violence. The Sentencing Project’s most recent data shows that roughly half of those serving federal sentences have been convicted of drug offenses. Analysis from the mid-2000s showed that a substantial portion of the federal prison population had no history of violence. These are not the dangerous offenders the SRA’s defenders invoked. They are people whose offense category, under the current structure, entitles them to no meaningful individualized review regardless of what they do while incarcerated.

The Longer Someone Stays in Custody, the More Exposure Accumulates

There is a liability dimension to this analysis that does not get enough attention. The federal government, through the Bureau of Prisons, assumes a constitutional duty of care for every person in federal custody. The Eighth Amendment prohibits cruel and unusual punishment. That prohibition, as interpreted by the courts, requires the government to provide adequate medical care, protect prisoners from known risks of serious harm, and maintain conditions that do not constitute deliberate indifference to basic human needs. The longer a person is in custody, the longer that duty runs, and the longer the window for something to go wrong.

The primary vehicle for claims against the federal government for custody-related injuries is the Federal Tort Claims Act. Under the FTCA, individuals can sue the United States for negligent or wrongful acts by federal employees acting within the scope of employment. Constitutional claims against individual BOP officers historically could be brought under the implied damages doctrine established in Bivens v. Six Unknown Federal Narcotics Agents, which the Supreme Court extended to Eighth Amendment medical care claims in Carlson v. Green in 1980.

The practical landscape is this: federal prisoners can bring FTCA claims for medical negligence and negligent custody. Bivens claims for constitutional violations by individual officers have become extraordinarily difficult to maintain since the Supreme Court’s 2017 decision in Ziglar v. Abbasi and the subsequent effective closure of new Bivens contexts. The government’s accountability structure for in-custody harm has narrowed precisely as the prison population has expanded and aged.

But the FTCA exposure is real, continuous, and proportional to the length of time someone remains in custody. The $700,000 settlement paid by the DOJ in 2024 for the death of Devon Gillians at FCI Coleman, after Bivens was unavailable and the claim was brought under the FTCA, illustrates the point. That is one settlement. The BOP manages roughly 155,000 people across 122 facilities, many of which are overcrowded, understaffed, and operating with documented deficiencies in medical care. The liability tail on each of those people runs from the moment they enter custody to the moment they leave. The policy that extends that period unnecessarily extends the liability window.

The aging of the federal prison population under a no-parole structure compounds this. Medical costs in federal prisons have increased dramatically. An aging incarcerated population presents with the full range of chronic conditions: diabetes, hypertension, cardiac disease, cancer. The constitutional mandate to provide care does not diminish because the care is expensive. The Bureau of Prisons is required, under Estelle v. Gamble and its progeny, to provide adequate medical treatment for serious medical needs. That obligation runs for the full duration of custody. Every year of unnecessary incarceration beyond what risk and rehabilitation analysis would support is a year of unnecessary medical liability.

What 2018 Fixed and What It Didn’t

The First Step Act of 2018 was a genuine bipartisan achievement. It expanded good time credits from 47 to 54 days per year of sentence imposed, applied retroactively. It created the earned time credit system that allows eligible prisoners to accrue credits toward prerelease custody through participation in recidivism reduction programming. It modified mandatory minimum sentences for certain drug offenses. It made the Fair Sentencing Act of 2010’s crack cocaine reforms retroactive. These were meaningful changes.

The data on recidivism outcomes is striking. Among the 44,000-plus people whose release was expedited by the First Step Act, the recidivism rate stands at approximately 9.7 percent, against a baseline of 46.2 percent for all BOP releases in 2018, per the Brennan Center’s June 2024 analysis of DOJ’s fourth annual report. That disparity does not tell us the First Step Act caused the difference, because the populations differ in important ways. But it is consistent with a substantial body of research showing that supervised, programmatic release tied to individual risk assessment reduces reincarceration.

What the First Step Act did not do is restore parole. It did not create individualized review of release eligibility. It did not modify the 85 percent floor for the majority of offenders. It did not address the 59,000 prisoners categorically excluded from earned time credits due to their offense of conviction. And it did not fix the PATTERN risk assessment tool’s documented problems with racial disparities and overestimation of recidivism risk, which determine which prisoners can access the most valuable credits in the first place.

The Council on Criminal Justice’s December 2024 report found that individuals released under the First Step Act in 2023 served 82.1 percent of their imposed terms, down from 89.7 percent pre-FSA. A 7.6 percentage point reduction. That is progress. It is not a solution to a structural problem that has compounded for 40 years.

What a Rational Policy Would Look Like

The argument for reforming the federal no-parole structure does not require the assumption that most federal prisoners are innocent or that sentencing guidelines were wrong in every case. The argument requires only three things the record already supports.

First: a significant portion of the federal prison population is serving sentences for non-violent offenses, primarily drug trafficking, at the 85 percent floor, with no meaningful individualized review of their risk or rehabilitation progress. The policy that keeps them incarcerated is not tied to any demonstrated public safety benefit; the First Step Act’s own recidivism data suggests supervised release for lower-risk offenders produces better outcomes at lower cost.

Second: the fiscal argument for the current system does not hold. Probation costs $3,433 per year. Federal prison costs $47,162. Community corrections cost $43,703. The government is paying roughly fourteen times more per person to warehouse non-violent offenders in federal facilities than it would cost to supervise them in the community. That gap was always going to produce pressure for reform; it is producing it now.

Third: the liability exposure generated by long-term federal custody of populations that could be safely supervised in the community is real, continuous, and proportional to the duration of incarceration. Medical obligations, safety obligations, the constitutional duty of care that attaches to every person in federal custody, all of these run for the full sentence. A system that extends sentences beyond what individualized risk assessment would support extends liability beyond what the policy’s public safety rationale can justify.

The Sentencing Reform Act was sold as fairness and fiscal responsibility. Four decades of data say it produced neither for the populations it most affected. The federal prison population multiplied sixfold before reforms began to slow it. The fiscal cost is measured in billions annually. The civil liability tail is documented and ongoing. And the recidivism data from the First Step Act’s limited reforms shows what the research consistently shows: supervised, individualized, programmatic release produces better outcomes at lower cost than the alternative.

The record on this is not complicated. The only complicated part is finding the political will to read it.