This study aims to determine the magnitude and characteristics of problematic cocaine consumption in the city of Barcelona, Spain. Capture-recapture with a single source was used to estimate prevalence. Log-linear regression models with interaction terms were fitted to the total sample and to subgroups according to other drugs consumed. Emergency room indicator data were obtained from the Barcelona Drug Information System. Drug-related emergencies of Barcelona residents for 1999 were analyzed. During 1999, a total of 4,035 drug-related emergencies were seen in Barcelona hospitals. Of these, 1,656 (41%) involved cocaine consumption; 41% of these patients had consumed cocaine with an opiate; 29% used cocaine with other substances; and 30% used cocaine alone. It was estimated that there was a total of 25,988 problematic cocaine users (95% confidence interval 11,782-58,064), yielding a rate of 31.27 per 1,000 inhabitants aged 15 to 54 years (95% confidence interval 14.2-69.9). The number of cocaine-related emergencies was high enough to allow capture-recapture to be applied, thus obtaining an estimate of the prevalence of problematic cocaine consumption, and high enough to characterize users according to different profiles. The use of capture-recapture with a single source can be interesting for problems related to the urban context.