The purpose of this study was to evaluate the impact of human immunodeficiency virus (HIV) infection on treatment for tuberculosis (TB). The study population comprised 28,522 black Southern African gold miners. Patients with sputum culture-positive new or recurrent pulmonary TB diagnosed in 1995 were prospectively enrolled in the cohort. Directly observed therapy (DOT) was practiced and outcomes were assessed at 6 mo after treatment was begun. There were 376 cases of TB (incidence 1,318 per 100,000), of which 190 (50%) were HIV positive and 82 (22%) had recurrent TB. There was no association between HIV status and history of previous TB or drug resistance. Neither the treatment interruption rate (2%) nor the rate at which patients transferred out of the treatment program (1.6%) were associated with HIV status. Excluding deaths, cure rates were similar for HIV-positive and HIV-negative patients (89% versus 88%), but significantly lower in those with recurrent than in those with new TB (77% versus 92%). Mortality was 0.5% in HIV-negative patients versus 13.7% in HIV-positive patients, and in the latter group was associated with CD4(+) lymphocyte depletion. Autopsy examination showed that in HIV-positive patients, early mortality was due to TB whereas late deaths were most commonly due to cryptococcal pneumonia. The study showed that a well-run TB control program can result in acceptable cure rates even in a population with a very high incidence of TB and HIV infection. Particular vigilance is needed for concurrent infections, which may contribute significantly to mortality during treatment of TB in HIV-positive patients.