Data for this study were collected as part of the Drug Use Forecasting (DUF) program. DUF interviews adult male and female arrestees four times each year in the police holdover facilities of 24 cities regarding their drug use, sex partners, and needle use. During the fourth quarter of 1990 and the first two quarters of 1991, researchers added items to interviews in the St. Louis DUF site to elicit information regarding subjects' knowledge of safe-sex practices, self-reported exposure to HIV, test experience for HIV and AIDS, condom use, and other steps to protect against HIV when having sex. Respondents were asked to describe their single most frequent or preferred safe-sex practice. The study shows that many, perhaps most, arrestees in the city of St. Louis engage in a variety of sex practices that place them at risk for contracting HIV. The prevalence estimate cannot be more precise because the study does not show what proportion of arrestees who reported no safe-sex practices engage in risky behavior. Large proportions of arrestees who practice safe sex use methods of questionable effectiveness. Owing to their heterogeneity, a single approach to prevention is not likely to be effective with arrestees. Multiple networks should be used to disseminate safe-sex information. Access to this high- risk population generally occurs only when they are under legal supervision. 1 figure and 28 references