There is now evidence from a wide variety of geographic areas that many intravenous drug users will change their behavior in order to reduce their risk of developing AIDS. There is even evidence from some areas that the behavior change has led to relative stabilization of seroprevalence rates, although longer-term studies will be needed to establish this definitively. AIDS behavior change in the area of sexual risk reduction appears to be much more difficult than change of drug injection behavior. Conceptual models of AIDS-related behavior change are needed, particularly models that can incorporate the injection of different drugs and variation in social and psychological characteristics among drug injectors. There is increasing evidence for a wider spectrum of HIV-related morbidity and mortality among intravenous drug users than is captured by the current surveillance definition for AIDS, again emphasizing the need for effective prevention programs.