We performed a meta-analysis and a decision analysis on the discontinuation of prophylaxis for Pneumocystis carinii pneumonia (PCP) in patients infected with human immunodeficiency virus who had adequate immune recovery while receiving highly active antiretroviral therapy. In the meta-analysis (14 studies with 3584 subjects who had discontinued prophylaxis), 8 cases of PCP occurred during 3449 person-years (0.23 cases per 100 person-years [95% confidence interval, 0.10-0.46]). In the decision analysis, mortality and time spent alive without immunodeficiency in the modeled discontinuation strategy were similar to those in the continuation strategy. For patients who received primary prophylaxis, the discontinuation strategy led to slightly fewer episodes of PCP and fewer toxicity-related prophylaxis withdrawals (e.g., 8.6 vs. 34.5 cases per 100 patients during a 10-year period). Patients on the discontinuation strategy were more likely to be receiving trimethoprim-sulfamethoxazole when they became immunodeficient. Comparative results were similar for patients with prior PCP. Discontinuation of PCP prophylaxis in patients with adequate immune recovery is a useful strategy that should be widely considered.