Parole & Sentencing

The Parole Release Pipeline: Two Decades of Contraction

The Georgia Board of Pardons and Paroles executed 10,164 total release actions in FY2001—7,305 paroles, 2,242 supervised reprieves, 615 conditional transfers, and 2 commutations (BOPP Annual Report FY2001 — Georgia Parole Board). In FY2002, 10,271 people were released to parole, with property, drug, and violent offense categories accounting for most releases (BOPP Annual Report FY2002 — Georgia Parole Board). The Board hit a modern peak of 13,374 releases through all forms of clemency in FY2016, then began a sharp downward slide: 10,468 in FY2017, 10,363 in FY2018, 6,245 in FY2022, 5,863 in FY2023, and 5,443 in FY2024 (BOPP Annual Reports FY2016–FY2024 — Georgia Parole Board). The FY2022 total alone was 2,389 fewer releases than the prior year, and FY2024 was 420 fewer than FY2023. FY2025 ticked up slightly to 5,588 (BOPP Annual Report FY2025 — Georgia Parole Board).

Measured against the FY2016 peak, the FY2025 total represents a decline of roughly 58 percent. The early totals include supervised reprieves and conditional transfers while recent reports describe parole-to-supervision releases, so the comparison is not perfectly like-for-like—but the direction is unmistakable. Georgia released about 45 percent fewer people through Board action in FY2025 than in FY2001 even though the state prison system remained above 50,000 people and the GDC amended FY2026 budget reached $1.799 billion (Georgia Department of Corrections Budget FY2026-FY2027).

The Shrinking Parole Docket and the Missing Denial Data

The release decline did not occur because Georgia ran out of parole-eligible people. The Board reported 24,738 individual parole-eligible cases considered in FY2019, 21,790 in FY2020, and 16,255 in FY2021—a 34 percent drop in two years (BOPP Annual Reports FY2019–FY2021 — Georgia Parole Board). Over the same period, total clemency votes cast by Board members fell from 86,054 in FY2019 to 75,522 in FY2020 and 59,354 in FY2021, a 31 percent decline. Because a single case can receive multiple votes, the vote total is not a headcount, but the two series tell the same story: the Board is making far fewer parole decisions than it did a few years earlier.

Later BOPP annual reports do not consistently publish a comparable cases-considered figure, leaving the post-FY2021 docket opaque. The collection Georgia's Parole System: Denial Rates, Life Sentences & Fiscal Impact frames parole outcomes alongside denial rates and life-sentence data, but GPS found that the Board's own reports do not present denial rates or demographic breakdowns with consistency. That gap is consequential: without knowing who is denied and on what basis, the public cannot evaluate whether high supervision-completion rates are the product of sound selection or of paroling only the lowest-risk applicants while people who are older, sicker, or serving long sentences remain inside.

Life Sentences and the Graying of Georgia's Prisons

Georgia's life-sentenced population was 10,392 people in 2024, according to a national census compiled in A Matter of Life: Life and Long-Term Imprisonment in the United States — Georgia in National Context (2024 Census). That includes 7,679 people serving life with the possibility of parole, 1,949 serving life without the possibility of parole, and 764 serving virtual life sentences of 50 years or more. One in five people in a Georgia prison—20 percent of the reported prison population—is serving a life or virtual life sentence.

These sentences are aging the system. Of 47,391 active inmates in the GPS database, 12,777—27 percent—are age 50 or older, and 8,694—18.3 percent—are 55 or older (Aging Prison Population & Compassionate Release: Georgia Data, National Research, Fiscal Analysis, and Legislative Landscape). National cost data show why that matters: CDCR estimated average annual health care costs of $87,552 for people in their 60s and $237,325 for those 80 and older in 2024. Georgia, meanwhile, had no hepatitis C test result on file for 27,471 people—51 percent of its June 2026 custody population—meaning the state lacks basic medical surveillance for a rapidly aging, long-sentenced population.

Sentencing Policy, Incarceration Rates, and Fiscal Consequence

Georgia's sentencing architecture has produced one of the deepest incarceration footprints in the country. The state holds 881 people per 100,000 residents—the 7th highest rate in the nation and higher than any country in the world except El Salvador (Recidivism & Reentry Failures in Georgia). Approximately 95,000 people are behind bars across Georgia's prisons, jails, immigration detention, and juvenile facilities, including roughly 53,000 in state prisons as of 2025; the Georgia Department of Corrections alone administers prison sentences for more than 50,000 people (Georgia Incarceration Trends: Population, Demographics & National Context; Georgia Department of Corrections Budget FY2026-FY2027). Another 191,000 people are serving felony probation—the largest felony probation population of any state (Georgia Probation & Community Supervision: Reform, Costs & Outcomes). In FY2002, Georgia's total correctional population was 194,485, made up of 126,659 people on probation, 46,814 inmates, and 21,012 parolees, showing that community supervision has long been the largest part of the footprint (BOPP Annual Report FY2002 — Georgia Parole Board).

Georgia's correctional footprint also has deep historical roots. In 1975, before the Earned Time Law, Georgia's institutions housed more than 11,000 people in facilities designed to hold 4,200, according to Dr. Allen Ault's public position paper cited in the first Earned Time System evaluation (The Earned Time System: First Annual Evaluation, NCJ 67770, 1977). The General Assembly enacted the Earned Time Law on July 1, 1976, specifically to address overcrowding, a negative subjective philosophy of corrections, and a lack of defined criteria for release decisions. The earlier Georgia Youthful Offender Act of 1972 had created an indeterminate sentence of zero to six years for non-capital offenders aged 17 through 24, with release consideration governed by a contract signed by the offender, an institutional representative, the Youthful Offender Division, and the Parole Board. That contract placed responsibility on the inmate to participate in specified vocational, educational, and counseling programs. Evaluators found that after 15 months as law, the Earned Time System was implemented "to a greater or lesser degree in every state and county institution," though successful elements coexisted with unsuccessful ones; field observations included counselor caseloads of 190 each at one Southwest District institution, while a county camp with similar caseloads but no managerial support completed plans sporadically and quarterlies and performance reports not at all.

By September 1, 1977, the total accountable population in Georgia's state and county correctional institutions and community facilities was 11,697, of whom 4,737—40 percent—were Earned Time inmates: 3,670 in state institutions (42 percent of the state institution population), 967 in county institutions (37 percent of the county institution population), and 100 in community facilities (32 percent of the community facility population). The first evaluation sampled 156 institutional files, representing 28 percent of the 568 Earned Time inmates permanently assigned longer than 60 days across the sampled institutions, and found the primary documentation uneven. Supervisor's Evaluations—the primary transaction document for periodic assessment—were present in only 40 percent of sampled files. Performance Plans were present in 80 percent of files, but only 34 percent were signed and dated by the Classification Committee. Quarterly Review Summaries were present in 70 percent of files, Intake Assessments in 90 percent, Extended Assessments in 80 percent, and Participation Status forms in 98 percent, with no sampled inmate having decided not to participate in ETS. Institutions developed 53 varied privilege lists, drawn from a master list of 68 distinct privileges. The privileges appearing most often across the 53 institutional lists were telephone calls (48 institutions), extra or extended visits (41), supervised fishing trips (23), detail changes (17), and days off work or activity (15). Across 11 institutions, evaluators recorded 1,396 privilege awards; telephone calls and visiting privileges accounted for 81 percent of the total, and the mean time between privilege award and redemption was 8.8 days. The distribution of those privileges was uneven: lower- and medium-security inmates comprised 44 percent of the sampled population but received 74 percent of the 755 privileges awarded, and white males comprised 38 percent of the sampled population but received 48 percent of all privileges. The average state institution population rose from 8,241 in FY76 to 8,640 in FY77, a 5 percent increase (The Earned Time System: First Annual Evaluation, NCJ 67770, 1977).

These earlier mechanisms show that Georgia has long tied release consideration to institutional performance—and that implementation gaps, documentation lapses, and uneven privilege distribution have been recurring issues.

A later baseline in Georgia Criminal Justice Data, 1988 (NCJ 128960, 1991) shows the corrections apparatus before the most recent growth. As of November 30, 1988, GDC was responsible for 18,870 inmates institutionalized in facilities under its authority—15,525 people in 25 state institutions and 2,924 in 27 county correctional institutions—and, as of December 1988, supervised 107,610 active probationers. The same report recorded that Georgia's county jails held more than 14,000 inmates in facilities designed for fewer than 10,000, including roughly 4,000 sentenced state inmates awaiting transfer to state institutions. As of November 1988, the Board of Pardons and Paroles employed 539 individuals, with 211 engaged in investigation and/or supervision of 10,101 parolees. The share of probationers on first offender status rose from 15.7 percent in 1984 to 17.0 percent in 1987, according to the same report. The state's 740 law enforcement agencies employed 17,196 certified officers, including 242 GBI special agents and 48 narcotics agents. The 1988 report recorded a statewide total of 397,129 index crimes, up from 358,842 in 1987, and 367,991 non-index crimes, up from 317,771 in 1987. Georgia law enforcement agencies made 67,002 index crime arrests in 1988, a 19.3 percent increase from 1987, including 17,773 violent index arrests and 49,229 property index arrests; within those totals, arrests included 707 murder, 1,102 forcible rape, 3,928 robbery, 12,036 aggravated assault, 10,493 burglary, 34,980 larceny, 3,441 motor vehicle theft, and 315 arson. County-level index crime rates ranged from 16,183.2 per 100,000 persons in Fulton County to 7,469.7 per 100,000 in Tift County among the top ten counties. At that time, GDC was authorized to employ 7,748 people, including 713 probation program employees, and operated 25 correctional institutions, 10 probation district offices, and 19 diversion centers; the diversion centers had an annual capacity of 3,451 residents and an average stay of 120 days, and the Department was authorized to employ 285 employees for diversion centers and 285 for detention centers, with the five detention centers expected to be fully operational by mid-1989 and an aggregate capacity of 220 beds serving 2,450 residents per year (Georgia Criminal Justice Data, 1988, NCJ 128960, 1991). These figures predate an era in which Georgia would come to hold roughly 53,000 people in state prisons and 191,000 on felony probation.

The system is also expanding financially. The GDC FY2026 original budget totaled $1.712 billion, and the Amended FY2026 budget reached $1.799 billion. Between January and May 2025, lawmakers approved roughly $634 million in new corrections spending—$434 million in the Amended FY2025 budget and $200 million in the FY2026 budget—the largest corrections funding increase in state history (Georgia's $600 Million Prison Spending Infusion: An Accountability Analysis). The collection Truth in Sentencing & Fiscal Impact: The $40 Billion Story places Georgia in a federal policy lineage: by 2001, 29 jurisdictions had received a combined $2.7 billion through the Violent Offender Incarceration and Truth-in-Sentencing grant program. Georgia is still paying for sentencing choices made decades ago.

Post-Conviction Remedies and the Limits of Legal Access

Georgia has created some back-end legal remedies, but their practical reach is limited. The Georgia Survivor Justice Act (HB 582), which creates resentencing rights for incarcerated domestic violence survivors, passed both chambers with overwhelming bipartisan support—only three dissenting votes total across both chambers (Georgia Survivor Justice Act (HB 582): Resentencing Rights, Legal Resources, and Support Organizations for Incarcerated DV Survivors). Yet GPS collections such as The Sleeping Giants: Two Georgia Statutes That Could Unlock Post-Conviction Justice and The People Behind the Case Law: Real Georgians Trapped by Judicial Narrowing of Post-Conviction Statutes document that courts have read existing post-conviction statutes narrowly, leaving people incarcerated despite legislative intent.

The historical record adds a caution about how release-related programming has operated in Georgia. The 1977 Earned Time System evaluation reviewed correctional behavior programs and concluded that such programs require voluntary, informed consent and client participation; it noted that some prison behavior programs had aimed at preserving institutional control rather than rehabilitation, and that in automatic good-time systems the removal of good time becomes an aversive control procedure. Justice model proponents in that evaluation argued that participation should be truly voluntary and not tied to release decisions. The same evaluation's file audit showed the practical meaning of that: Supervisor's Evaluations were present in only 40 percent of sampled files, and only 34 percent of Performance Plans were signed and dated by the Classification Committee (The Earned Time System: First Annual Evaluation, NCJ 67770, 1977). That historical insight underscores the narrowness of current post-conviction remedies: legal access and documentation quality have never been neutral in Georgia's release decisions.

National data on compassionate release illustrate how much legal representation alone can determine outcomes. Under Illinois' Joe Coleman Medical Release Act, 60 percent of applicants with legal counsel were granted release, compared with just 19 percent of those without lawyers. In 2022, Illinois denied 50 of 78 medical release applications—just under two-thirds—even under a statute designed for compassionate release (Aging Prison Population & Compassionate Release: Georgia Data, National Research, Fiscal Analysis, and Legislative Landscape). In Georgia, where the aging prison population is growing and post-conviction relief remains narrow, the same dynamic raises a hard question: if release depends on legal resources most prisoners do not have, how many statutory remedies are functioning as intended?

The Success-Rate Paradox: High Parole Completion, High Re-Arrest

Georgia's parole supervision outcomes appear strong in isolation. In FY2025, 73 percent of parolees successfully completed supervision, compared with a national average of about 60 percent. The pattern persisted for two decades: 60 percent in FY2005 against a national average of 46 percent; 66 percent in FY2009 against 45 percent; 74 percent in FY2013 against 53 percent; 70 percent in FY2023 against 57 percent; and 72 percent in FY2024 against 60 percent (BOPP Annual Reports FY2005–FY2025 — Georgia Parole Board). The Board has also emphasized swift enforcement: in FY02, the average time between a reported parole violation and a sanction was one day, and the Board's earlier research found that a one-percent drop in recidivism would save taxpayers $6 million annually (BOPP Annual Report FY2002 — Georgia Parole Board).

But completion of parole supervision is not the same as long-term desistance. The Bureau of Justice Statistics found that 76.6 percent of state prisoners released in 30 states in 2005 were rearrested within 5 years, and 83 percent were rearrested within 9 years (A Sense of Purpose as a Driver of Rehabilitation in Incarcerated People: An Evidence Brief). GPS collections on reentry and rehabilitation argue that the state's release and supervision model is not structured to produce durable success.

Georgia-specific recidivism forecasting data reinforce that distinction. The NIJ Recidivism Forecasting Challenge used an aggregated dataset of roughly 26,000 people released from Georgia prisons on discretionary parole for post-incarceration supervision in 2013; recidivism was measured as a new felony or misdemeanor arrest episode within three years of the parole supervision start date (Predicting Criminal Recidivism Using Specialized Feature Engineering and XGBoost (NIJ Recidivism Forecasting Challenge, Georgia parolees released 2013) (NCJ 305039, 2021)). The authors' XGBoost model performed best in 10-fold validation—mean Brier score 0.0945 and F1 score 0.8559—compared with Random Forest (0.1087, 0.8390) and a neural network (0.1525, 0.7850). Out-of-sample Year 1, Year 2, and Year 3 Brier scores were 0.1837, 0.1172, and 0.0720, with F1 scores of 0.3009, 0.2512, and 0.0492. The authors noted that all models suffered from heavy class imbalance because non-recidivism cases outnumbered recidivism cases, and they recommended the F1 score for future imbalanced-data studies. They also identified COMPAS as one of the earliest criminal justice AI tools and argued that their models can supplement existing recidivism technologies. The top Year 1 predictors were Age_at_Release, Prior_Arrest_Episodes_Felony, Gang_Affiliated, Prison_Years, and Prior_Arrest_Episodes_Property, while Year 2 and Year 3 models relied most on Recidivism_Arrest_PrevYear, Percent_Days_Employed, Jobs_Per_Year, Age_at_Release, and Avg_Days_per_DrugTest. In the challenge's hidden-data results, Oracle led Year 2 female-parolees with a Brier score of 0.1233, followed by MCHawks (0.1242), VT-ISE (0.1260), and DEAP (0.1263); for Year 2 combined male and female parolees, MCHawks led at 0.1405, ahead of Oracle (0.1451), VT-ISE (0.1472), and DEAP (0.1481).

An earlier controlled evaluation of Georgia's Cognitive Skills Program told a similar cautionary story about completion and selection. Phase I randomly assigned 468 parolees across 16 parole districts to a treatment group (N=232) or a control group (N=236). The overall 4 percent reduction in returns to prison and re-arrests for the experimental group was not statistically significant, but for participants who completed the cognitive skills classes, discrete-time survival analysis controlling for background characteristics produced statistically significant treatment effects of 17 percent on returns to prison and 19 percent on re-arrests relative to the control group. Phase II expanded to 1,193 offenders randomly assigned between July 1998 and April 2000 at 28 sites—25 parole districts and three correctional institutions—with 609 in the treatment group and 584 in the control group (The Georgia Cognitive Skills Experiment: Process Evaluation, Phase Two, NCJ 248529, 2001).

The Phase II experimental cohort was high-need. It was 91.3 percent male (545 males and 52 females among 597 experimental group members with sex recorded), 68.5 percent nonwhite—of the 409 nonwhite participants, one was Native American and the rest were African American—and had an average age of 32.2, with 69.0 percent between ages 18 and 35. Nearly two-thirds, 65.4 percent, had not completed high school, and 26.4 percent scored below a fifth-grade reading level on the WRAT reading assessment. The average IQ was 100.8, but 8.2 percent had IQs below 80. Half, 50.9 percent, were not employed full time at arrest; 56.8 percent had a history of substance abuse; 58.1 percent had multiple prior incarcerations; and 44.1 percent had prior parole revocations. Prior convictions were pervasive: only one participant, 0.2 percent, had no prior felony convictions, while 48 percent had three or more. Property convictions were the most common at 72.4 percent, followed by drug possession at 62.0 percent, violent convictions at 43.9 percent, and drug sale convictions at 42.0 percent. The study classified 49.2 percent as medium/high risk and 50.8 percent as low risk (The Georgia Cognitive Skills Experiment: Process Evaluation, Phase Two, NCJ 248529, 2001).

Phase II also documented implementation and data limitations. The 52 randomized program groups were supplemented by 10 additional classes across eight sites (N=98) that fell outside the experimental design because parole officers selected participants for those classes. Eighteen sites conducted two Cognitive Skills courses, another eighteen sites conducted three, and Whitworth Detention Center held four groups. Complete social and demographic data were available for 91.1 percent of the randomly assigned sample (N=1,087), and criminal history data were available for 98.2 percent (N=1,171). Among the experimental group, Jesness Inventories were available for 541 participants, 88.8 percent; 82 inventories, 15.2 percent, posed scoring problems due to missing or double responses, 56 of which could be fixed through scoring rules, leaving a final total of 513, or 84.2 percent, that generated personality types (The Georgia Cognitive Skills Experiment: Process Evaluation, Phase Two, NCJ 248529, 2001).

Georgia's first Earned Time System evaluation made a related point in 1977: it cited findings that for many inmates the payoff of time off the end of a sentence may be too far away to act as a reinforcer, while proximate reinforcers have meaning even to people serving long sentences. The evaluation documented how proximate reinforcers worked in practice: across 11 institutions, telephone calls and extra visits accounted for 81 percent of the 1,396 privileges recorded, and the mean time from privilege award to redemption was 8.8 days (The Earned Time System: First Annual Evaluation, NCJ 67770, 1977). That should caution against treating program completion or supervision completion as durable rehabilitation. If the parole docket continues to shrink while the state pours a record $634 million into new corrections spending, the high completion rate may simply reflect a smaller, lower-risk release cohort—not a system that is preparing people to leave prison for good.