Solitary Confinement

Scale, Duration, and the Georgia SMU

Solitary confinement in the United States is practiced at a scale difficult to fully account for. Estimates from 2014 placed the national population in isolation at 80,000–100,000; by 2016, the first Liman Center census counted approximately 68,000; the 2018 ASCA-Liman Nationwide Survey found 49,197 individuals — 4.5% of the population across 43 reporting prison systems — in restrictive housing, projected to approximately 61,000 nationwide. By 2019, 31,542 people were documented in restrictive housing across 39 reporting states — representing 3.8% of the total prisoner population — though Solitary Watch and Unlock the Box, drawing on BJS and Vera data and a survey of jails, estimated approximately 122,000 people in restrictive housing across prisons and jails combined, roughly 6% of the total incarcerated population. The 2021 estimate ranged between 41,000 and 48,000, with researchers noting that pandemic-era lockdowns may have expanded use significantly. These numbers should be understood as floors, not ceilings: reporting is inconsistent, definitions vary by jurisdiction, and many states do not disclose data voluntarily. In 36 jurisdictions reporting on duration, 25 counted more than 3,500 individuals held in restrictive housing for more than three years. (Solitary Confinement & Restrictive Housing)

Georgia's own record within this national pattern is stark. As of May 2026, GDC houses approximately 53,571 incarcerated people, with an additional 2,372 individuals backlogged in county jails awaiting transfer — figures that supersede the U.S. Department of Justice's October 2024 findings letter, which documented "almost 50,000" people in custody across 34 state-operated and 4 private prisons. The state's primary isolation unit — the Special Management Unit (SMU) at Georgia Diagnostic and Classification Prison — was designed for approximately 192 single-bunked cells and housed approximately 180 people at the time of the 2017 Haney inspection. As of July 2017, 78% of SMU prisoners — 141 of 182 people — had been in isolation for more than two years. (Solitary Confinement & Restrictive Housing) Approximately 20% of those held in the SMU had been confined there for six or more years, and the average duration of confinement was three to four years. By contrast, the 2019 national census found that 46% of people in restrictive housing had been held for three months or less, suggesting Georgia's SMU represented an extreme tail of long-term isolation even by national standards.

The human cost of these durations is visible in individual cases. Timothy Gumm — the lead plaintiff in Gumm v. Ford — was held in the SMU continuously for more than seven years, from 2010 to 2017, before being transferred. Robert Watkins, an additional named plaintiff, had been held for at least seven years at the time of the 2018 amended complaint. Daniel Barfield had been confined in the SMU for eight years at the time of the 2017 SCHR letter. These are not outliers in the statistical sense — they are the predictable product of a system that, as Gumm litigation documented, released residents directly from isolation to the community at sentence expiration, without any transitional programming. More recently, GPS Case #40 documents Christian Yandel Flores Tirado (GDC# 1003554733), a confirmed MH-3 prisoner held in segregation at GDCP and Rutledge State Prison with documented emotional deterioration — illustrating that long-term isolation of mentally ill people in Georgia's restrictive housing units remains an ongoing practice, not a closed chapter.

The SMU also housed a population with significant mental health needs. Dr. Craig Haney's 2017 inspection found that 70 of the SMU's 180 inmates — approximately 39% — were designated as mentally ill, and that it was "dangerous" to house them under those conditions. That figure almost certainly understates true prevalence, given the documented failures of mental health screening and classification.

Reforms That Have Reduced Solitary Confinement and Improved Safety Elsewhere

Evidence from multiple states and nations demonstrates that deep reductions in solitary confinement are achievable without jeopardizing safety, and that alternative models can improve institutional climate and reduce violence.

Remarkable reductions in isolation populations. California’s Ashker v. Brown settlement moved more than 1,512 individuals out of solitary and reduced Pelican Bay’s long-term isolation population from 513 to 2 — a 99.6% drop — with no reported surge in violence. Statewide, California’s total SHU/solitary population fell 65% between December 2012 (9,870) and August 2016 (3,471), continuing to decline to 594 by June 2018. In North Dakota, solitary confinement use fell 74.28% between 2016 and 2020; at one facility (JRCC) the monthly rate of solitary sanctions dropped 99%, and at the State Penitentiary (NDSP) it fell 59.1%.

Targeted programming reduces disciplinary infractions and violence. In Oregon, the Resource Team — a multidisciplinary intervention for individuals with extensive solitary histories (averaging 9.7 prior admissions) — cut disciplinary infractions by 55.7% and assaults by 73.9% among participants with at least three interactions. Oregon’s Behavioral Health Unit saw an 86% drop in staff use-of-force from 2016 to 2021. Pennsylvania’s “Little Scandinavia” unit at SCI Chester, which operates with a 1:8 officer-to-resident ratio (compared to 1:128 in the rest of the facility), has recorded only one incident categorized as violent since opening and almost no violence in 2024. The unit was renovated for approximately $300,000–$310,000, with a per-inmate daily cost roughly 1.5 times that of double-celling. In March 2025, Pennsylvania’s DOC announced expansion of the model to three additional facilities.

International comparisons show that high-staff, treatment-oriented models yield low recidivism. Norway’s prison system, designed for a 1:1.1 staff-to-inmate ratio, spends $127,671–$129,222 per prisoner per year. Its reconviction rate is 18% within two years and 25% within five years, down from 60–70% before reforms. The low-security Bastøy Island prison, with ~115 inmates and about 69–72 staff (only 3–5 overnight), has a recidivism rate of 16%. Norway has cut its prison rate by approximately 46% from 2005 to 2016 and closed roughly half its prisons while crime fell.

Decarceration without crime increases is widely documented. New York more than halved its prison population from 1999 to 2023 while its violent crime rate fell 34%, faster than the national 28% decline. This pattern — decarceration without crime increases — is described as the best-documented finding in the broader evidence base, with additional examples from New Jersey, California, and the Netherlands.

Cost-conscious oversight and staffing innovations are feasible. New Jersey operates a fully independent corrections ombudsperson office for about $2.8 million per year with 26 staff, reporting to the governor. Pennsylvania reduced its correctional-officer vacancy rate from 10.5% to 4.8% in two years by creating a dedicated recruitment division. Solitary confinement itself is noted to require twice as many guards, making its reduction a potential staffing efficiency.

Cautions on cross-jurisdictional comparison. Nordic and Western European outcomes reflect much shorter sentences, far smaller prison populations, and stronger social-welfare baselines. They establish that measurable alternatives exist — not that those models are directly transferable without adaptation. Recidivism definitions (rearrest, reconviction, reincarceration) and follow-up periods vary across jurisdictions and are specified where sources provided them. All figures are attributed to their source contexts.