Respondents were classified into one of seven groups, based on the total number of unprotected heterosexual acts reported in the prior 30 days and at follow-up. Resulting risk groups were compared on a series of demographic, economic, health, drug treatment, and drug use variables. Risk categories included decrease, increase, remaining at low risk, remaining at high risk, and no sexual activity. Bivariate and multivariate analysis indicated that, compared to persons at high levels of unprotected sex at follow-up, those who remained at a low level or were at a decreased risk level at follow-up were more likely to be HIV positive. Age, living alone, and having a stable source of income were also significant predictors of risk pattern. Risk pattern was not associated with type of risk reduction intervention (standard or enhanced) or with drug treatment (yes or no) between baseline and follow-up. Implications of the findings are discussed with respect to the efficacy of AIDS prevention interventions and the analysis of risk behavior changes over time. 22 references and 5 tables