Retrospective studies suggest that the mortality rate from HIV-1-associated tuberculosis is greater than that from tuberculosis alone, but it is not clear if this is due to failure of antituberculosis treatment or to the complications of HIV-1 infection. We have carried out a prospective cohort study of patients with tuberculosis in Nairobi, Kenya, to compare mortality rates, risk factors, and causes of death in HIV-1 positive and HIV-1 negative patients. One hundred seven HIV-1 positive and 174 HIV-1 negative patients with tuberculosis attending two tuberculosis treatment centers in Nairobi were enrolled and followed monthly. Mortality was significantly higher in HIV-1 positive than in HIV-1 negative patients within 6 months of the start of antituberculosis treatment after adjustment for age, sex, and education (rate ratio = 3.8; 95% confidence interval, 1.7 to 8.1; p less than 0.001). Most of the excess mortality occurred after the first month of treatment and was due to nontuberculous infections. Predictors for mortality differed greatly between HIV-1 positive and HIV-1 negative patients. Mortality was greater in HIV-1 positive patients treated with a "standard" regimen for tuberculosis than in HIV-1 positive patients receiving a "short-course" regimen (p = 0.08 when adjusted for all independent risk factors). Tuberculosis control programs in developing countries need to implement "short-course" regimens and train health workers to recognize and treat nontuberculous infections to maintain their effectiveness in the face of the HIV epidemic.