The optimal regimen for treatment of Mycobacterium avium complex (MAC) disease has not been established. Eighty-five AIDS patients with disseminated MAC disease were randomized to receive a three-drug regimen of clarithromycin, rifabutin or clofazimine, and ethambutol. Two dosages of clarithromycin, 500 or 1,000 mg twice daily (b.i.d.), were compared. The Data and Safety Monitoring Board recommended discontinuation of the clarithromycin dosage comparison and continuation of the rifabutin vs. clofazimine comparison. After a mean follow-up of 4.5 months, 10 (22%) of 45 patients receiving clarithromycin at 500 mg b.i.d. had died (70 deaths per 100 person-years) compared with 17 (43%) of 40 patients receiving clarithromycin at 1,000 mg b.i.d. (158 deaths per 100 person-years) (relative risk, 2.43; 95% confidence interval, 1.11-5.34; P = .02). After 10.4 months, 20 (49%) of 41 patients receiving rifabutin had died (81 deaths per 100 person-years) compared with 23 (52%) of 44 patients receiving clofazimine (94 deaths per 100 person-years) (relative risk, 1.20; 95% confidence interval, 0.65-2.19; P = .56). Bacteriologic outcomes were similar among treatment groups. In treating MAC disease in AIDS patients, the maximum dose of clarithromycin should be 500 mg b.i.d.