We examined the severity of immune deficiency in patients with HIV-associated tuberculosis in Côte d'Ivoire and assessed its effect on mortality and response to treatment. Consecutive patients attending a tuberculosis treatment centre in Abidjan with smear-positive pulmonary or clinically diagnosed extrapulmonary tuberculosis were tested for HIV-1 and HIV-2 infections and had CD4 lymphocyte counts measured. Patients received standard short-course chemotherapy. Analysis of outcome (restricted to smear-positive tuberculosis patients) was done at 6 months. The 247 HIV-positive patients were significantly more likely than the 312 HIV-negative patients to have CD4 lymphocyte counts of less than 200/microL (43% vs 1%; odds ratio 56.9; [95% CI 19.7-185.3]) and 200-499/microL (39% vs 14%, odds ratio 3.8; [2.5-5.9]). Among HIV-positive patients, median CD4 lymphocyte counts in those with extrapulmonary tuberculosis (198/microL; n = 67) was lower, but not significantly so, than among those with pulmonary tuberculosis (257/microL; n = 180). Among 460 patients with pulmonary tuberculosis, the overall mortality rate was significantly higher in HIV-positive than HIV-negative persons (6% vs 0.4%; relative risk 17.1 [2.2-131.4]), and increased with the severity of immune deficiency; mortality rates in HIV-positive patients with CD4 counts of < 200/microL and 200-499/microL were 10% and 4%, relative risk 27.6 (3.5-220.8); and 11.5 (1.2-109), respectively, compared to HIV-negatives. Among patients completing treatment, cure rates were similar in HIV-positive patients (93%) and HIV-negative patients (92%), and were not related to CD4 counts. Severity of immune deficiency was the major determinant of mortality in HIV-associated tuberculosis. Among people completing treatment, microbiological response was satisfactory irrespective of serological or immune status.