The purpose of this study was to compare the efficacy and safety of dapsone and trimethoprim/sulfamethoxazole in the primary prophylaxis ofPneumocystis cariniipneumonia (PCP) in patients infected with the human immunodeficiency virus (HIV) and having <200 CD4-positive cells per ml. This was a prospective, randomized, open-label study, using dapsone (100 mg p.o.) or trimethoprim/sulfamethoxazole (160 mg/800 mg p.o.) daily. Patients who developed toxicity requiring discontinuation were offered to cross over to the other study drug. They continued in the study until development of toxicity or documented PCP. Eighty-six patients were enrolled; 47 were randomized to receive dapsone and 39 to receive trimethoprim/sulfamethoxazole. Discontinuation of initial study drug occurred in 33 of the dapsone group and 25 of the trimethoprim/sulfamethoxazole group. Rash was the most common reason for discontinuation. Ten patients crossed over from dapsone to trimethoprim/ sulfamethoxazole (4 successfully), and 11 patients crossed over from trimethoprim/sulfamethoxazole to dapsone (6 successfully). During 1,638 patient-months of observation (862 for dapsone and 776 for trimethoprim/sulfamethoxazole), one episode of PCP developed in each group. Both dapsone and trimethoprim/sulfamethoxazole are efficacious for the prophylaxis of PCP in HIV-infected persons with <200 CD4-positive cells per ml, but are each associated with significant toxicity. Development of toxicity to one dlrug does not invariably predict toxicity to the other.