We report the results of a randomized multicenter study comparing the efficacy of antithymocyte globulin (ATG) with that of cyclosporin A (CsA) as first-line therapy for severe aplastic anemia (SAA). Patients were randomized to receive ATG and prednisone (PDN) or CsA; hematologic response and toxicity were compared. At 3-month evaluation, patients who had no or minimal response received the alternative therapy to assess the value of a sequential immunosuppressive therapy for treatment of severe aplastic anemia. One hundred nineteen patients were randomized; 25 were excluded, of whom 3 were misdiagnosed and 22 did not follow the cross-over protocol. Ninety-four patients were analyzed; 46 received CsA, and 48 received ATG-PDN. The actuarial survival was 66.7%, with a median follow-up time of 19 months. There was no significant difference in survival between the groups with, at 3 months, an actuarial survival of 88% in the CsA group and 75% in the ATG group (NS); at 12 months, it was 70% in the CsA group and 64% in the ATG group (NS). The percentage of complete and partial response was 11.6% and 16%, respectively, at 3 months, and 31.6% and 30%, respectively, at 12 months (NS). The main prognostic factor was the absolute neutrophil count (ANC) at entry: Patients with ANC less than 0.2 x 10(9)/L had a significantly lower survival as compared with patients with more than 0.2 x 10(9)/L ANC (P = .0001). At 12 months, 62 evaluable patients were alive, with a complete or partial response in 36 patients. Patients who had responded to the first treatment had a better recovery of bone marrow failure than those who had sequential immunosuppression. The main complication was infection, which was more often observed and more often lethal during ATG and PDN therapy. In this study, initial treatment of SAA with either CsA or ATG-PDN followed by cross-over therapy for nonresponders produced comparable response and survival rates.