We developed a decision-analytic model to assess the effectiveness and costs of dapsone, trimethoprim-sulfamethoxazole, or aerosolized pentam-idine as initial prophylaxis forPneumocystis cariniipneumonia in human immunodeficiency virus-infected people without prior symptoms AIDS and with CD4 counts <200/mm3. Each strategy increased life expectancy by about 18% compared with no prophylaxis; annual per-person costs were $440, $700, and $1,680 for dapsone, trimethoprim-sulfamethoxazole, and aerosolized pentam-idine, respectively. These cost differences make a strategy beginning with dapsone most cost effective, with an incremental cost-effectiveness ratio of $13,400 per life year saved compared with no prophylaxis. Aerosolized pent-amidine was substantially less cost effective, but the incremental cost effectiveness ratios were highly dependent on estimates for quality of life, efficacy, toxicity, and compliance. We conclude that, based on currently available data, initial prophylaxis with either dapsone or trimethoprim-sulfamethoxazole is most cost effective. For every 100,000 people treated, starting prophylaxis with trimethoprim-sulfamethoxazole or dapsone—with crossover to aerosolized pentamidine if oral therapy is not tolerated—may save between $98 million and $124 million per year.