To estimate HIV incidence and identify predictors of seroconversion among Montreal street youth. Methods: From 1995 to 2000, street youth aged 14-25 years were recruited in a prospective cohort study. Interviews were conducted semiannually and included anti-HIV antibody testing. Among subjects who tested HIV negative at study entry and were interviewed at least twice, predictors of HIV seroconversion were identified using Cox regression. Variables considered as potential predictors were age, sex, injection drug use, being a male reporting male sexual partners, and survival sex. Results: Overall, 1013 youth were recruited in the study. HIV prevalence at study entry was 1.4% [95% confidence interval (CI) 0.8-2.4] and was stable over the 6 recruitment years. Among the 863 subjects selected for the incidence analysis, 66.7% were boys, 47.2% had ever injected drugs at study entry, and 25.7% had ever engaged in survival sex. The selected participants cumulated 2327 person-years of follow-up and 16 HIV seroconversions were observed, for an incidence rate of 0.69 per 100 person-years (95% CI 0.39-1.11). In univariate analysis, injection drug use [hazard ratio (HR), 7.0] and involvement in survival sex (HR, 4.0) were associated with HIV incidence. In the multivariate analysis, only injection drug use was retained. Conclusions: Among Montreal street youth, injection drug use was the strongest predictor of HIV seroconversion. Prevention of initiation into injection drug use must become a public health priority. (C) 2003 Lippincott Williams Wilkins.