Recidivism is a person’s return to criminal behavior after they finish a sentence, complete probation, or leave custody. It is the single concept that drives most debates over sentencing length, prison programming, parole conditions, and reentry support, because the question of whether punishment actually stops future crime depends on how often it fails to. Bureau of Justice Statistics data tracking people released from state prisons found that an estimated 83% were re-arrested within nine years of release.1Bureau of Justice Statistics. 2018 Update on Prisoner Recidivism: A 9-Year Follow-Up Period (2005-2014) That figure is startling on its own, but it only means something once you know what counts as recidivism, who gets counted, and what the law does in response.
The Three Ways Recidivism Gets Measured
There is no single official definition. Researchers and agencies use three different benchmarks, and the one they pick changes the number dramatically.
Re-arrest is the broadest. It counts any new arrest by law enforcement after release, whether or not charges are ever filed and whether or not a court finds the person guilty. Because arrest requires only probable cause, this measure sweeps in the widest range of behavior and produces the highest figures. It is the standard metric in national studies.2Bureau of Justice Statistics. Recidivism and Reentry
Re-conviction is stricter. It requires a court to find the person guilty of a new offense, which means the case cleared the full weight of due process. Many arrests end in dropped charges, plea reductions, or acquittals, so re-conviction rates always run lower than re-arrest rates for the same group of people.
Re-incarceration is the narrowest. It counts only people who actually return to jail or prison, whether for a new felony conviction or for violating the terms of their supervision. This threshold is the hardest to meet and produces the lowest numbers, but it captures the most serious failures of reintegration.
Technical Violations Count Too
A large share of re-incarceration has nothing to do with a new crime. Technical violations of supervised release or parole, such as failing a drug test, missing an appointment with a probation officer, refusing mandated treatment, or possessing contraband, can trigger revocation proceedings on their own.3United States Courts. Just the Facts: Revocations for Failure to Comply with Supervision Conditions and Sentencing Outcomes
In the federal system, the maximum prison term a court can impose on revocation depends on the seriousness of the original conviction. A person originally convicted of a Class A felony faces up to five years on revocation. The cap drops to three years for a Class B felony, two years for a Class C or D felony, and one year for anything less serious.4Office of the Law Revision Counsel. 18 USC 3583 – Inclusion of a Term of Supervised Release After Imprisonment Those caps apply whether the violation involves a new offense or not.
What the National Numbers Look Like
The most comprehensive federal data comes from a BJS study tracking more than 400,000 people released from state prisons in 34 states in 2005. Within three years, an estimated 68% had been arrested for a new crime. By six years, that figure climbed to 79%. By nine years, 83% had been re-arrested at least once.1Bureau of Justice Statistics. 2018 Update on Prisoner Recidivism: A 9-Year Follow-Up Period (2005-2014)
Those numbers deserve context. Re-arrest sweeps in every encounter with law enforcement, including minor offenses and arrests that never result in charges. Re-conviction and re-incarceration rates for the same group are substantially lower, though BJS has noted that data limitations in some states make precise national figures for those benchmarks harder to calculate.5Bureau of Justice Statistics. Recidivism of State Prisoners
The steepest climb happens in the first year after release. That early window is when a person is most likely to run into the triggers that restart criminal behavior: housing instability, unemployment, severed social ties, and the supervision conditions that themselves generate new legal contact. By around year five, the rate of new arrests begins to level off.
Who Is Most Likely to Reoffend
Researchers have identified a short list of variables that reliably predict risk.
Criminal history is the strongest predictor. The more prior arrests and convictions a person has, the higher their statistical risk of a new offense. A long, varied record presents a very different profile from a first-time offender.
Age is the other dominant factor. People who first enter the justice system as teenagers or young adults reoffend at significantly higher rates than those whose first arrest comes later in life. Criminal behavior tends to peak in the late teens and early twenties and declines with age. The PATTERN risk assessment tool used in federal prisons assigns its highest risk points to inmates aged 18 to 25 and zero points to those over 60.
Socioeconomic conditions fill out the picture. Lacking a high school diploma or GED correlates with higher recidivism, partly because it narrows employment options. Unemployment is itself a strong predictor. So is housing instability. Someone released with no job, no diploma, and no stable address is statistically far more likely to be re-arrested than someone with even one of those anchors in place. These factors do not cause reoffending in any direct way, but they describe the conditions under which it becomes much more likely.
How Courts and Corrections Estimate Risk
Courts and corrections agencies increasingly use algorithmic tools to estimate the likelihood that a person will reoffend. The tools take the predictors above, assign weights, and produce a score that places the person in a category such as low, medium, or high risk. Those scores feed into pretrial release decisions, sentencing, prison programming, and parole.
In the federal pretrial system, United States probation officers use the Pretrial Risk Assessment, or PTRA, to help decide whether a defendant should be released or detained before trial. The PTRA estimates the risk of failing to appear, committing new crimes, and violating pretrial conditions.6United States Courts. Pretrial Risk Assessment Officers use it alongside a full investigation, not as a standalone verdict.
For federal inmates, the First Step Act of 2018 required the Attorney General to develop a risk and needs assessment system. The resulting tool, PATTERN, classifies inmates as minimum, low, medium, or high risk for both general and violent recidivism.7Office of the Law Revision Counsel. 18 USC 3632 – Development of Risk and Needs Assessment System PATTERN mixes static factors that cannot change, such as criminal history score and whether the current offense was violent, with dynamic factors that can, such as program participation, disciplinary infractions, drug treatment completion, and education level. The dynamic factors are the leverage point: completing programs and avoiding infractions can lower a score and open the door to earlier release.
Accuracy and Bias Concerns
These tools consistently outperform unstructured human judgment at predicting recidivism, but outperforming is not the same as being accurate. A widely publicized analysis of one state-level tool found it correctly identified future offenders only about 61% of the time across all crime types, and just 20% of the time for violent crime specifically. The same analysis raised concerns that the tool falsely flagged Black defendants as high-risk at nearly twice the rate of white defendants.8United States Courts. False Positives, False Negatives, and False Analyses
Subsequent peer-reviewed research challenged those findings on methodological grounds, concluding that when the tool was tested using established statistical standards for bias, a given score translated into roughly the same likelihood of reoffending regardless of the defendant’s race.8United States Courts. False Positives, False Negatives, and False Analyses The debate is unresolved. What is clear is that these tools are decision-support instruments, not crystal balls.
How the Law Punishes Repeat Offenders
Every jurisdiction in the United States has some form of habitual offender law that increases punishment for people with repeat convictions. The logic is simple: if standard sentences did not deter someone, progressively harsher penalties should.
The most visible version is the three-strikes model, adopted in various forms by many states. Under a typical three-strikes law, a person convicted of a third qualifying serious or violent felony faces a mandatory sentence of 25 years to life. Which offenses qualify, how many strikes are required, and the resulting sentences vary widely by state, but the underlying concept is consistent.
The Armed Career Criminal Act
At the federal level, the Armed Career Criminal Act is one of the most consequential recidivism statutes. Under the ACCA, a person convicted of illegal firearm possession who has three prior convictions for violent felonies or serious drug offenses, committed on separate occasions, faces a mandatory minimum of 15 years in prison, up to a maximum of life.9Office of the Law Revision Counsel. 18 USC 924 – Penalties Without the enhancement, the same firearm charge carries a 10-year maximum. The jump from a 10-year ceiling to a 15-year floor is enormous, and courts cannot suspend the sentence or grant probation.
In 2024, the Supreme Court held that the Sixth Amendment requires a jury, not a judge, to determine whether prior offenses were committed on separate occasions before the ACCA enhancement can apply. Prosecutors must now prove that element to the jury beyond a reasonable doubt.
The Federal Safety Valve
Congress has carved out a narrow exception to federal mandatory minimums. Under 18 U.S.C. ยง 3553(f), a court can sentence below a mandatory minimum if the defendant’s criminal history falls within strict limits: no more than four criminal history points (excluding one-point offenses), no prior three-point offense, and no prior two-point violent offense, all calculated under the federal sentencing guidelines.10Office of the Law Revision Counsel. 18 USC 3553 – Imposition of a Sentence The defendant must also have used no violence, must not have led the offense, and must have cooperated with the government by providing all available information about the crime. The safety valve is the clearest acknowledgment in federal law that not every person facing a mandatory minimum belongs in the same sentencing box.
Programs Designed to Reduce Recidivism
The First Step Act of 2018 was the most significant federal sentencing and prison reform legislation in a generation. Its premise is that reducing recidivism requires more than punishment. The law directed the Bureau of Prisons to expand access to evidence-based programming and created tangible incentives to participate.
Under the Act, federal inmates earn 10 days of time credits for every 30 days of successful participation in approved recidivism reduction programs or productive activities. Inmates classified as minimum or low risk who maintain that classification over two consecutive assessments earn an additional five days per 30-day period, for a total of 15.7Office of the Law Revision Counsel. 18 USC 3632 – Development of Risk and Needs Assessment System Credits can be applied toward early transfer to prerelease custody, such as a halfway house or home confinement, or toward early placement on supervised release.
Qualifying programs include cognitive behavioral therapy, drug treatment, vocational training, education, and work programming. Research from the National Institute of Justice has found that cognitive behavioral therapy programs in prison settings are effective or promising at reducing reoffending roughly 74% of the time when rigorously evaluated, and that therapeutic approaches outperform punishment-based interventions.11National Institute of Justice. Does Cognitive Behavioral Therapy Work in Criminal Justice?
The Collateral Consequences That Feed the Cycle
The formal sentence is only part of the picture. People with criminal records, especially repeat convictions, face a web of collateral consequences that make reintegration harder and reoffending more likely. These consequences are rarely announced at sentencing, but they can last a lifetime.
Public Benefits
Federal law imposes a lifetime ban on SNAP and TANF eligibility for anyone convicted of a felony involving possession, use, or distribution of a controlled substance. The ban applies only to drug felonies. States can opt out or limit it by passing their own legislation, and many have done so to varying degrees. In states that have not, a person released from prison after a drug conviction may be permanently ineligible for basic food assistance, and the household’s benefit amount is reduced by treating the ineligible person as if they do not exist for calculation purposes.12Office of the Law Revision Counsel. 21 USC 862a – Denial of Assistance and Benefits for Certain Drug-Related Convictions
Social Security benefits follow different rules. If you are convicted and incarcerated for more than 30 continuous days, Social Security disability benefits are suspended for the duration of confinement and can be reinstated the month after release. Supplemental Security Income works similarly, but with a harsher cliff: incarceration of 12 consecutive months or longer terminates SSI eligibility entirely, requiring a new application after release.13Social Security Administration. Benefits After Incarceration: What You Need To Know
Employment
Criminal records create major barriers to work, which is itself one of the strongest predictors of whether a person will reoffend. For federal jobs, the Fair Chance to Compete for Jobs Act prohibits federal agencies and contractors from asking about criminal history before making a conditional offer.14U.S. Department of the Interior. Fair Chance to Compete Act Exceptions exist for positions requiring security clearances, law enforcement roles, and certain military positions. Many states and cities have passed similar “ban the box” laws covering private employers, though the scope varies.
Housing
Public housing authorities screen applicants for criminal history, and certain convictions can mean denial of admission or eviction. Federal law requires exclusion of anyone subject to lifetime sex offender registration and anyone convicted of manufacturing methamphetamine in federally assisted housing. Beyond those mandates, housing authorities have broad discretion to set their own policies, so access to affordable housing after release depends heavily on local decisions.
Taken together, these consequences are part of what makes recidivism such a difficult problem. The same factors that predict reoffending (unstable housing, lack of employment, no access to benefits) are often made worse by the conviction itself. That is the loop the measurement, the risk scores, the sentencing enhancements, and the reduction programs are all trying, in different ways, to break.