In France a harm-reduction policy was implemented in the late 1980s with the aim of reducing the prevalence of HIV and hepatitis C virus (HCV) infection among drug users. The ANRS-Coquelicot survey was designed to measure the prevalence of HIV and HCV infection among drug users and to examine determinants of at-risk behaviors. In 2002, information was collected from 166 drug users recruited in all types of services specializing in drug use intervention and harm reduction in Marseille, France. Self-reported HIV and HCV serostatus was compared with the results of serological tests done on capillary blood collected on filter paper. The self-reported and biologically documented prevalence rates of HIV infection were identical (22%). In contrast, the self-reported prevalence of HCV infection was 52%, whereas the biologically documented prevalence was 73%. Overall, 30% of HCV-infected drug users were unaware of their status. Forty-four percent of drug users under 30 years of age were HCV seropositive, suggesting that they had been infected early during drug use. The harm-reduction policy seems to have had a marked impact on HIV transmission among drug users but a much more limited impact on HCV transmission. The limitations and implications of the study are discussed