To assess the incidence of infection with HIV-1 in the Maryland state prison system, inmates with excess sera stored from specified intake periods between April and June 1985 and 1986 were approached in May 1987 to volunteer for venipuncture. Of the 2286 inmates for whom intake specimens were stored, 1038 (45.4%) no longer incarcerated as of April 1987 were excluded from the study; another 319 missed the survey. Of the 929 eligible inmates approached for the study, 446 (48%) consented and 422 (94.6%) provided a specimen. Twenty-nine (6.6%) were confirmed seropositive at time of entry into prison, indicating that infection had occurred prior to incarceration. Baseline seropositives were more likely (P less than 0.05) to be non-violent offenders, committed in Baltimore City, and black. The 393 participants seronegative at baseline provided a total of 482 prison-years of potential exposure to infection. Two inmates seroconverted with baseline specimens seronegative on Western blot and follow-up sera confirmed positive; their duration of pre-incarceration detention was 69 and 146 days, respectively. No interruption of incarceration was recorded for these two inmates. The rate of infection in this prison sample, which does not include an average of 2 months of pre-incarceration detention for the study sample, was 0.41% per prison-year.