In an attempt to compare the efficacy of rifampin-doxycycline with tetracycline-streptomycin for the treatment of human brucellosis, we administered both combinations for a 30-day period, similar to the period recommended by the World Health Organization in a prospective, randomized trial. Forty-six patients were included in the final study (36 men and 10 women); 41 had blood cultures positive for Brucella melitensis. The 28 patients in group A received tetracycline hydrochloride at doses of 0.5 g every 6 h or doxycycline at 100 mg every 12 h for 30 days plus 1 g of streptomycin a day for 21 days. The 18 patients in group B received rifampin at 15 mg/kg per day in a single morning dose plus 100 mg of doxycycline every 12 h for 30 days. For patients with focal disease from both groups, therapy was prolonged to 45 days. All patients underwent rigorous clinical and bacteriological long-term follow-up. There were no therapeutic failures in either group, and the defervescence period was similar for both groups (3.1 days for group A, 2.6 days for group B). Two patients (7.1%) from group A had relapses, as did seven (38.8%) from group B (P = 0.024), and blood cultures again became positive for B. melitensis in all of them. In both groups treatment was generally well tolerated. The results strongly suggest that the rifampin-doxycycline combination is a less efficacious mode of therapy for brucellosis to prevent relapses than is the classical tetracycline-streptomycin combination when both are administered for 30 days. A more prolonged period of administration of the rifampin-doxycycline combination may be required to obtain the same low relapse rate as that achieved with the classical tetracycline-streptomycin treatment.