Scores Without Sanitation: Why Georgia's Prison Food-Safety Numbers Don't Reflect What Inmates Eat From

On paper, Georgia's prisons run clean kitchens. A scan of the Georgia Department of Public Health's food-service inspection database — the same system that scores restaurants, hospitals, and schools — shows GDC facilities posting overwhelmingly high marks. Most score in the 90s. Many hit perfect 100s. A handful slip to a B or C. Three have failed outright since 2022. That distribution would suggest a food-service system that is, at worst, uneven and, at best, well-run.

The problem is that the inspection score measures something narrower than the public believes it measures. A DPH grade reflects what a sanitarian observed during one announced walkthrough on one day in one kitchen. It does not measure what arrives on the tray. It does not measure what happens between inspections. And it does not measure what the incarcerated workers who clean those kitchens — who lift the serving tables, who pull bulk goods from the warehouse, who pour cornmeal from sacks — see when the inspector is not there. This page is about that gap.

The Distribution: Mostly A's, Sharp Exceptions

GPS reviewed roughly 240 DPH inspections of GDC facilities from mid-2023 through April 2026 and found that the vast majority cluster at the top of the scale. Central State Prison scored a perfect 100 in November 2024, June 2025, and again in November 2025. Baldwin scored 100 in April 2024, June 2025, and December 2025. Hancock recorded a string of 100s across 2023, 2024, and 2025. Calhoun, Lee, Walker, Rutledge, Riverbend, Dooly, Long Unit, McRae Women's, Hays, Coffee, and Georgia Diagnostic and Classification State Prison all posted scores in the high 90s or perfect 100s during 2025.

Against that backdrop, the failures stand out. Three GDC kitchens have scored below the 70-point passing threshold since 2022: Johnson State Prison received a 64 on December 11, 2023 — the lowest documented score in the dataset — followed by a 67 on a December 20, 2023 follow-up; Pulaski State Prison scored 67 on January 29, 2026; and Smith State Prison sat in a long trough of B's and C's, including a 72 in February 2026 and an 85 in June 2025. A handful of other facilities — Coastal, Telfair, Dodge, Macon — have drifted into B and C territory on individual visits without crossing the failure line.

The headline number, then, is that something like 95% of inspected GDC kitchens pass on any given visit, and a small set repeatedly fails. The thesis of this page is that the headline number is the wrong number.

What a Failure Actually Looks Like

The three failing inspections are worth reading closely because they describe conditions that do not develop overnight and do not vanish overnight.

At Johnson State Prison's December 2023 inspection — the 64 — the inspector documented multiple rats and roaches throughout the kitchen and noted that the problem was ongoing "with little to no change." Bulk food items including oil, flour, and rice bran had holes gnawed through their bags, with visible rat droppings and urine on the packaging. Multiple cold-holding foods registered above 41°F and were discarded on the spot. Five cooking ovens, a tilting skillet, a cooking kettle, a griddle, a freezer unit, and the bulk ice machine were all broken. Holes were found in floors, walls, and ceilings. A follow-up on December 20, 2023 brought the score up only to 67 — still failing. By March 2024, Johnson scored an 86 (B); by December 2024 it had climbed back to a 96.

Pulaski State Prison's January 2026 failure tells a similar story of conditions that had been visible for months. The facility's only designated handwashing sink was nonfunctional — the plumbing had been ripped from the wall, the pipe smashed inward. Sewage was backing up through floor drains, flagged as a repeat violation. Nacho meat registered 65°F and a sauce at 123°F, both hot-holding violations also flagged as repeats. Employees were observed switching between tasks without washing their hands. Looking back across the prior year, Pulaski's trajectory was already declining: 83 in February 2025, 73 in August 2025, 78 on a September 2025 follow-up, then the failing 67 in January 2026. A follow-up eight days later produced a 96.

At Smith State Prison, rodent activity was noted in every single inspection from 2022 through 2025. The February 2026 routine inspection — which scored 72 — documented roach activity in the bakery and tray-making station (repeat), broken handwashing sinks, broken plumbing in four sinks, a broken dishwasher faucet, clogged floor drains, damaged walls, and mildew on walls, floors, and ceilings (repeat). The May 2022 inspection score of 68 was a failing grade. Nearly four years later, Smith had moved from a 68 to a 72.

The pattern that emerges across all three facilities is not "the system caught a problem." It is "the system documented the same problem, repeatedly, and the problem remained." Pulaski's sewage. Smith's rodents. Johnson's roaches. The violations carry the word repeat on official DPH forms because the conditions persist between inspections. That word is the part of the inspection regime that does not work.

What the Score Does Not Measure

The inspection score is a snapshot. It captures, with reasonable fidelity, what a sanitarian could see during a single visit, usually scheduled in advance because of security protocols. As The Marshall Project reported in March 2025, health departments generally must arrange prison inspections ahead of time, and even when inspectors find serious violations, they are reluctant to shut down a prison kitchen because the people who eat from it have nowhere else to go.

A scheduled visit has two predictable consequences. First, kitchens prepare for it. Surfaces are wiped down, traps are checked, hot-holding equipment is adjusted, and incarcerated workers — who do nearly all of the labor — are pulled to clean. The system the inspector observes on inspection day is not necessarily the system that runs the other 364 days. Second, the inspection assesses kitchen sanitation, not tray quality. It evaluates whether cold holds are below 41°F and hot holds above 135°F at the moment of measurement; it does not evaluate whether the protein on the tray is what the menu says it is, whether portions match the contracted weights, or whether the meal arriving at a housing unit fifty yards away has been kept warm in transit.

GDC's own Standard Operating Procedure 409.04.26, titled "Food Service Permits-Health Department Inspections," confirms the structure: prison kitchens obtain permits from and undergo inspections by county health departments under O.C.G.A. § 26-2-370, which explicitly extends food-service jurisdiction to "institutions, both public and private." A separate GDC rule, Board of Corrections Rule 125-1-2-.10, mandates internal "regularly scheduled and unannounced evaluative inspections" by GDC staff. The DPH records are publicly accessible. The internal GDC inspections are not. The only inspection layer the public can see is the announced one.

Testimony From Inside the Same Kitchens

The clearest evidence that the score and the kitchen are different objects comes from the people who work inside them. GPS has collected firsthand accounts from incarcerated kitchen workers and longtime residents at facilities whose inspection scores would suggest clean operations.

A Tell My Story author identified as Stony, who has been in the Georgia system since 2015, described the kitchens at Jackson — the same Georgia Diagnostic and Classification State Prison that DPH scored a perfect 100 on March 26, 2025 — in stark terms: "The first thing you notice at Jackson is the roaches. They were everywhere. On the bottoms of the trays, and because trays are stacked, that meant they were on the tops of trays too. They scattered when you set your tray down on the table. Too often they were in the food itself — sometimes dead, sometimes still alive." That observation is dated to a period in which the same kitchen was registering A-grade inspections.

Carla Simmons, writing in Truthout from a Georgia women's facility, described "roach legs in cornbread and rats climbing over the dry goods in the warehouse." An incarcerated writer publishing in Filter Magazine as "Jimmy Iakovos" relayed a kitchen worker's account of being told to "shake the spoon" to short portions, and described inmates whose "teeth are loose, bodies gray and bony" as a consequence of the food. None of these accounts surfaces in an inspection score. None of them would change the score if it did, because none describes the kind of point-in-time violation the inspection form captures.

GPS's own reporting on Telfair, Smith, Johnson, and Pulaski has compiled additional accounts from kitchen-detail workers describing roach-lined serving tables, mice traveling through dry storage, and droppings inside bulk-good containers — at facilities whose DPH scores range from the high 80s to perfect 100s. The contradiction does not require interpretation. A kitchen that consistently scores 91 and a kitchen in which incarcerated workers find mouse droppings in flour can be the same kitchen, because the inspection does not look for the second condition on the days it is looking for the first.

The Small-County Structure

The third reason the score is a weak instrument is structural. Several of Georgia's prisons are located in counties where the prison is among the largest employers and where the local public-health district inspects a small universe of food-service establishments. The sanitarian who inspects a prison kitchen in a rural county may also inspect the same county's schools, restaurants, and nursing homes, and may work in an office whose budget, staffing, and political relationships are intertwined with the county's largest institution. This is not an accusation against any individual inspector — the DPH inspectors named in the database appear to be doing the job as designed. It is an observation about the design.

A regulatory regime in which the inspecting agency is geographically and economically adjacent to the inspected facility, and in which inspections are scheduled rather than surprise, and in which the inspected facility has no realistic alternative provider if shut down, is a regime in which the cost of finding serious violations is high and the cost of not finding them is low. The pattern of repeat violations at Smith, Pulaski, and Johnson — conditions documented across multiple visits without producing structural change — is what such a regime produces. Inspectors note the problem. The problem remains. The next inspection notes it again.

The Litigation Backdrop

Federal court records add a layer the inspection database does not capture, with one important caveat. In Gumm v. Ford (Case No. 5:15-CV-41, M.D. Ga.), the lead plaintiff alleged that food served in the Special Management Unit at Georgia Diagnostic and Classification State Prison was "inedible or have less nutritional content," undercooked, cold, rotten, or beyond expiration date. He lost 40 pounds. A class settlement was approved in May 2019 requiring SMU prisoners to receive the same food access as the general population. The case is relevant here because GDCP's main kitchen — the kitchen DPH inspects — is generally not the food chain that reaches the SMU. SMU trays are typically assembled and routed separately, which means the kitchen the inspector visits and the food the SMU receives are partially decoupled. A perfect inspection score at the main kitchen does not speak to what arrives at the SMU door, and the Gumm record suggests the two were measuring different realities even as both existed at the same facility.

The U.S. Department of Justice's October 2024 findings report, drawn from a three-year investigation of 17 GDC prisons, documented food deprivation as part of the totality of conditions it found unconstitutional under the Eighth Amendment — including the case of a man at Calhoun State Prison who died of dehydration and renal failure after his cell door flap was locked shut and no meals were delivered for days. Calhoun's kitchen, in the same period, scored consistently in the high 90s. The kitchen was clean. The food chain to the cell was not. The DPH inspection is not designed to detect that distinction.

What the Score Is Good For, and What It Isn't

None of this is to say the DPH inspection database is useless. It is the only piece of regular, externally administered, publicly accessible oversight that touches the GDC food system. It catches the worst kitchens — the Johnsons, the Pulaskis, the Smiths — and it documents the repeat-violation pattern that makes the failure visible. Compared to states whose prisons self-inspect with no external authority, Georgia's regime is meaningfully better.

What the database is not good for is the question most members of the public would ask of it: Is the food my tax dollars are buying being prepared safely and arriving on the tray in edible condition? The score answers a narrower question — Did the kitchen meet sanitation thresholds on the day a sanitarian visited? — and the answer to the narrower question is often "yes" at facilities where the answer to the broader question is "no." A 100 at Jackson and roaches in the tray stack are not in contradiction. They describe two different measurements of the same place.

The reform implication is not that DPH inspectors should be replaced or accused. It is that an inspection regime built for restaurants, in which the customer can walk out and the operator can be shut down, does not transfer cleanly to a setting in which the customer is locked in and the operator cannot be shut down. Unannounced visits, public access to GDC's internal Rule 125-1-2-.10 inspections, an independent food-service oversight body of the kind Massachusetts has proposed in Bill H.4125, and structural separation between the inspecting authority and the local political economy of the inspected facility would each address part of the gap. The DPH score alone will not.

Sources

This analysis draws on roughly 240 DPH food-service inspection reports for GDC facilities from 2023 through April 2026, retrieved from the Georgia Department of Public Health's public inspection database; the U.S. Department of Justice's October 2024 findings report on Georgia prison conditions; federal court filings in Gumm v. Ford (M.D. Ga.); GDC Standard Operating Procedures including 409.04.26, 409.04.13, and 409.04.28; Board of Corrections Rule 125-1-2-.10; reporting by The Marshall Project, the Atlanta Journal-Constitution, Filter Magazine, and Truthout; and firsthand accounts from incarcerated kitchen workers and longtime GDC residents collected through Georgia Prisoners' Speak's Tell My Story project.