We report the results of the first prospective randomized multicenter study of immunosuppressive treatment in patients with previously untreated nonsevere aplastic anemia (AA) as defined by a neutrophil count of at least 0.5 x 10(9)/L and transfusion dependence. Patients were randomized to receive cyclosporin (CSA) alone or the combination of horse antithymocyte globulin ([ATG] Lymphoglobuline; Merieux, Lyon, France) and CSA. The endpoint of the study was the hematologic response at 6 months. One hundred fifteen patients were randomized and assessable with a median follow-up period of 36 months; 61 received CSA and 54 ATG and CSA. In the CSA group, the percentage of complete and partial responders was 23% and 23%, respectively, for an overall response rate of 46%. A significantly higher overall response rate of 74% was found in the ATG and CSA group, with 57% complete and 17% partial responders (P =. 02). Compared with CSA alone, the combination of ATG and CSA resulted in a significantly higher median hemoglobin level and platelet count at 6 months. Fewer patients required a second course of treatment before 6 months due to a nonresponse. In the CSA group, 15 of 61 (25%) patients required a course of ATG before 6 months because of disease progression, compared with only 3 of 54 (6%) in the ATG and CSA group. The survival probabilities for the two groups were comparable, 93% (CSA group) and 91% (ATG and CSA group), but at 180 days, the prevalence of patients surviving free of transfusions, which excluded patients requiring second treatment because of nonresponse, death, disease progression, or relapse, was 67% in the CSA group and 90% in the ATG and CSA group (P =.001). We conclude that the combination of ATG and CSA is superior to CSA alone in terms of the hematologic response, the quality of response, and early mortality, and a second course of immunosuppression is less frequently required.