The objective was to compare the efficacy and tolerance of monthly aerosolized pentamidine versus trimethoprim-sulfamethoxazole (TMP-SMX) to prevent the first episode of Pneumocystis carinii pneumonia (PCP) in human immunodeficiency virus (HIV)-infected patients. In an open, prospective, randomized multicentric clinical trial, HIV-infected patients (n = 214) with CD4 cell counts < 200/mm3 or 20% without a history of PCP or cerebral toxoplasmosis were randomized to receive for at least 2 years aerosolized pentamidine (300 mg monthly) or low-dose daily TMP-SMX (400-80 mg). The mean follow-up was 578 days. The two groups (except for gender) were homogeneous for age, risk group for HIV infection, initial CD4+ lymphocyte count, and mean follow-up. The PCP rate per year of observation using an intent-to-treat analysis was 3.1% and 1.3% in the groups treated with pentamidine and TMP-SMX, respectively (p > 0.05). Moderate or severe clinical and biological side effects were observed in five patients on pentamidine and 33 on TMP-SMX (p < 0.05). Nineteen episodes of cerebral toxoplasmosis were diagnosed during the study. The analysis showed no significant difference in time of development of toxoplasmosis, but only one patient was actually treated with TMP-SMX. Survival was not significantly different in the two groups. Low-dose daily TMP-SMX or monthly aerosolized pentamidine effectively prevented a first episode of PCP in HIV-infected patients, but aerosolized pentamidine was better tolerated. However, TMP-SMX is less costly and should have a preventive effect for toxoplasmosis.