Aims. To determine the health status of a cohort of heroin addicts. Design. Longitudinal follow-up study of a cohort for 10.5 years (March/July 1985-December 1995) based on official registers. Setting. Catalonia, Spain. Participants. One hundred and thirty-five heroin addicts who started treatment in four specialized drug addiction services in Catalonia in 1985. Measurements. (a) Total and annual numbers of AIDS and tuberculosis diagnoses and admissions to drug treatment; (b) annual morbidity rate; (c) average annual morbidity rate; and (d) standardized morbidity ratio. Kaplan-Meier (log rank test) was used to assess predictive factors. Findings. During the follow-up period, 34 heroin addicts were diagnosed with AIDS (25%), the average annual morbidity rate was 2.7% and the standardized morbidity ratio was 82. The most frequent diseases indicative of AIDS were Pneumocystis carinii pneumonia (11 cases) and tuberculosis (seven cases of disseminated or extrapulmonary tuberculosis and two cases of pulmonary tuberculosis). Twenty addicts were diagnosed with tuberculosis and 15 were co-infected with AIDS and tuberculosis. Seventy-six addicts were readmitted to treatment and generated 124 drug treatment episodes, 94 of whom were drug-free and 30 on methadone maintenance. Conclusions. Follow-up through registers is a novel, low-cost technique that may provide important and comparable information on the long-term evolution of drug addicts in different European regions. The measures of disease frequency and association obtained allow us to describe and compare the distribution of patterns of infectious complications (AIDS and tuberculosis) and relapses in such a pervasive disorder as heroin addiction. Other advantages of follow-up based on official registers are efficiency, reliability, sensibility and comparability.