The proposal requests 30 months of support to continue findings of the first study of significant associations between the structure of social networks of intravenous drug users (IVDUs) and their HIV risk behavior and to study the stability of IVDU networks. Preliminary data indicates that the intervention with IVDU drug sharing networks as a group was successful in producing reduction in risky behavior compared to randomly assigned time matched control in a short term follow-up. A cross over design will be used to measure durability of the effect in a longer time interval potency of the intervention with the previously untreated randomly assigned control group. Issues of validity of self-reports of risky behavior are addressed in this study with an experimental method involving the use of peer informants who report on the subjects pattern of risky behavior.