OBJECTIVE: To assess the impact of tuberculosis on mortality in patients with AIDS. DESIGN: Community based cohort study. SETTING: 52 centres in 17 countries (AIDS in Europe study). SUBJECTS: 5249 patients who were alive and free of tuberculosis one month after the diagnosis of AIDS, enrolled between 1979 and 1989, and followed up until 1992. MAIN OUTCOME MEASURES: Onset of clinically active tuberculosis or death, or both. RESULTS: During a mean follow up period of 15 months 201 (4%) patients developed tuberculosis and 3889 (74%) died. Patients who developed tuberculosis survived significantly longer (median 22 months) than those who did not (median 16 months). This apparent survival advantage was due to patients who survived longer having more opportunity to develop tuberculosis (or any other disease). In models that took into account the time at which tuberculosis was diagnosed, the onset of tuberculosis was associated with a significant increase in mortality (adjusted relative hazard of death 1.34; 95% confidence interval 1.12 to 1.60). CONCLUSIONS: The onset of tuberculosis in patients with AIDS predicts a substantial increase in mortality. Whether this increased mortality is directly attributable to the tuberculosis remains uncertain. If the association is causal preventive chemotherapy and aggressive treatment of tuberculosis could improve survival in AIDS.