Objective:To compare the efficacy of two- and three-drug regimens for treating Mycobacterium aviumcomplex (MAC) bacteremia in patients with AIDS.Design:Randomized open-label clinical trial.Setting:Outpatient HIV specialty centers' clinics.Patients:A total of 106 adults with AIDS and MAC bacteremia.Interventions:Patients were treated with clarithromycin 500 mg twice daily and ethambutol 800–1000 mg daily and were randomized to receive clofazimine 100 mg daily or no clofazimine.Main outcome measures:Quantitative blood MAC cultures, symptoms, adverse reactions and survival.Results:Patients randomly assigned to three drugs had significantly higher baseline colony counts of MAC in blood than patients receiving two drugs. The proportion of patients becoming culture-negative was 65% in the two-drug group and 54% in the three-drug group. The median time to negative culture was 58 days for patients in the two-drug group and 63 days for the three-drug group. At the last visit during treatment, the mean reduction in colony forming units/ml of MAC in blood was 1.8 log10for the two-drug group and 2.3 log10for the three-drug group. Improvement in fever and night sweats was reported by 87 and 89% of the two-drug patients and 84 and 86% of the three-drug patients. During the study, 38% of two- drug patients and 61% of three-drug patients died (P= 0.032), and time to death was shorter in patients treated with three drugs (P = 0.012). In a multivariate analysis, both assignment to clofazimine and high baseline colony counts of MAC bacteremia were significantly associated with death (P< 0.05).Conclusion:The addition of clofazimine to a regimen of clarithromycin and ethambutol for MAC bacteremia in AIDS patients does not contribute to clinical response and is associated with higher mortality.