To assess the cause and clinical severity of diabetic ketoacidosis in male prisoners hospitalized in New York City. Retrospective chart review. A municipal hospital in New York City. Forty-nine adult male prisoners with a total of 54 hospital admissions for diabetic ketoacidosis between January 1, 1989, and June 30, 1991. Charts were reviewed for diabetic and medical history, time from arrest until hospitalization, cause of diabetic ketoacidosis, admission laboratory data, and hospital course. Thirty-eight (70%) of the 54 admissions for diabetic ketoacidosis among prisoners occurred because prisoners had not received insulin during the period immediately following arrest (mean number of days from arrest until hospitalization was 2.5). All of these individuals had a history of insulin-dependent diabetes and were reportedly compliant with their insulin regimen at the time of arrest. Admission laboratory data for this group of prisoners included a mean serum glucose level of 27.4 mmol/L (495 mg/dL) and a mean serum bicarbonate level of 14.4 mmol/L. Mean number of days in the hospital was 3.4 including a mean of 1 day in an intensive care unit. Inadequate access to medication results in serious sequelae for recently arrested prisoners in New York City with insulin-dependent diabetes. Access to health care for recently arrested prisoners needs to be improved.