Despite widespread use of trimethoprim-sulfamethoxazole (TMP-SMZ) for prophylaxis in neutropenic patients, questions remain regarding its efficacy, toxicity, the risk of selection of resistant isolates, and the relation of its activity to selective decolonization vs. the attainment of direct inhibitory levels within blood and tissues. We evaluated the effect of TMP-SMZ (160/800 mg orally every 12 hours) in 42 adult granulocytopenic patients (< 100 absolute neutrophils/mm3, mean duration 13.3 days) undergoing chemotherapy for acute leukemia at 11 participating Veterans Administration Medical Centers in a randomized, double-blind, placebo-controlled trial. No significant differences in survival, frequency of bacteremia, overall infections, use of systemic antimicrobial therapy, or adverse effects, including myelosuppression, were observed between patients receiving TMP-SMZ vs. those receiving placebo. All patients acquired trimethoprim-resistant organisms. Concentrations of trimethoprim in serum were significantly lower before febrile episodes than when patients were afebrile. These results suggest that the purported activity of TMP-SMZ may be related to the serum concentration achieved. Moreover, the results highlight the need for additional study of the value of antibiotic prophylaxis in neutropenic patients.