Hepatitis due to anti-tuberculosis therapy is an infrequent, but potentially devastating event. In HIV positive patients with tuberculosis (TB), the consequences are likely to be even greater, as they frequently require other hepatotoxic medications. The object of our study was to determine the frequency of toxic hepatitis during therapy for TB. Included were 198 patients with a presumed or confirmed diagnosis of tuberculosis; of whom, 69 were HIV positive (35%), 75 were negative (38%) and 54 had unknown HIV status (27%). Toxic hepatitis occurred in 15/198 (8%) patients. The incidence of hepatitis in HIV patients was much greater than in HIV negative/unknown [RR=7.5 (2.2-25.6); p=0.0001] and the onset of hepatitis was short (median 7 days in HIV patients). During TB therapy, 1 in S (12.5%) patients taking ketoconazole developed hepatitis; 9/53 (17%) taking sulfamethoxazole-trimethoprim [RR=3.4 (1.1-9.3); p=0.03]. Among the 15 patients who developed hepatitis 11 required hospitalization (mean 19 days), 5 died (33.3%), 2/15 (13%) due to hepatitis. HIV positive patients had a significantly higher rate of toxic hepatitis during anti-tuberculosis therapy than those without HIV infection. Hepatitis occurred just after initiation of TB treatment. Clinical findings were non-specific and hepatic enzyme elevations were moderate, yet hospitalization and mortality rates were high. This suggests that in settings where careful monitoring of patients early in their course of TB treatment is routine, morbidity and mortality may be low, but poor monitoring would have potentially serious consequences. There is a need for new drug treatments (schedules or regimens) for TB in an effort to reduce these adverse events.