Twenty-one episodes of Pneumocystis carinii pneumonia (PCP) and the acquired immune deficiency syndrome (AIDS) were treated with corticosteroids in the form of intravenous methylprednisolone or oral prednisone. A standard dose of 80 mg/day x 5 days was given for 15 episodes, whereas 6 patients received variable doses of 20-120 mg/day x 4-20 days. All were treated with trimethoprim-sulfamethoxazole (TMP-SMX). Comparison was made with 12 AIDS patients with PCP who were not treated with steroids. The steroid group was more severely ill than the controls as measured by alveolar-arterial oxygen difference but were otherwise comparable. Mortality from the pneumonia in the steroid group was 2/21 (10%) vs. 3/12 (25%) in the control group. Significant differences were seen in the following parameters: time to defervescence (1 day vs. greater than 9.3 days), the proportion of patients with pO2 greater than 70 mm Hg at day 5 [12/21 (57%) vs. 1/12 (9%)] and at day 10 [19/21 (90%) vs. 7/12 (58%)], and number of adverse drug reactions [4/21 (19%) vs. 9/12 (75%)]. There were fewer late relapses [1/19 (5%) vs. 2/9 (22%)] after a 5.5 month (mean) follow-up. All patients had improvements in their clinical status when initially given corticosteroid therapy but early relapses occurred when steroids were discontinued in five patients (24%). No other complications could be attributed to steroid therapy in this study. A brief course of high-dose corticosteroids appears to be beneficial in severely ill AIDS patients with pneumocystis pneumonia. This suggests the need for randomized, double-blind, placebo-controlled trials to confirm these findings.