To assess the effectiveness of two regimens with allopurinol or pentavalent antimony as secondary prophylaxis for visceral leishmaniasis (VL) in human immunodeficiency virus (HIV)-infected patients. DESIGN: Retrospective, nonrandomized, open trial. SETTING: A 1,000-bed academic tertiary institutional hospital in Barcelona. PATIENTS: Forty-six individuals over 14 years old with HIV infection, who recovered from an episode of VL between January 1988 and February 1995. INTERVENTIONS: Twenty patients did not receive any prophylaxis, nine received 300 mg/8 h of allopurinol, and 17 received 850 mg once-a-month of pentavalent antimony. Patients were followed-up every 3 months, and the endpoint of study was relapse of VL. RESULTS: Twenty-one patients had recurrent VL: 13 of 20 in the control group (65%), 5 of 9 in the allopurinol group (56%), and 3 of 17 in the antimonial group (18%). Kaplan-Meier estimates of the probability of remaining relapse-free at 12 months were 9% without prophylaxis (95% Cl, 0-22%), 21% with allopurinol (95% Cl, 0-51%), and 93% with antimonials (95% Cl, 82-100%) (P < 0.001). Multivariate analysis showed that the only significant variables related to relapsing course of VL were assignment to the antimonial group, and the fact that the patient had experienced a previous episode of VL. CONCLUSIONS. Pentavalent antimony given once a month is effective in the prevention of VL relapses in HIV-infected individuals. It is a low-cost treatment that proved to be well tolerated. Therefore, pentavalent antimony should be considered a suitable agent for secondary prophylaxis against VL.