The outcome of 155 adult aplastic anemia (AA) patients treated with antithymocyte globulin (ATG, Upjohn, Kalamazoo, MI) at University of California, Los Angeles from 1977 to 1988 was evaluated. The median survival of the 146 patients who did not undergo bone marrow transplantation was 5.6 years, with 49% +/- 4% surviving more than 6 years. The most important predictor of survival was positive response to ATG (P < 0.001), which was observed in 48% of patients. Among pretreatment variables, disease severity was the best predictor of survival. Patients with moderate AA (MAA) had significantly better survival than those with severe (SAA) or very severe (VSAA) disease (P = 0.04). The 6-year actuarial survival rates of the three groups were 71% +/- 9%, 48% +/- 7% and 38% +/- 7%, respectively. Cox regression analysis found disease severity to be the only pretreatment variable significantly associated with survival (P = .02). Patient age, sex, disease etiology, concurrent treatment with androgens, or duration of ATG therapy were not associated with differences in survival or response to ATG. Late clonal hematologic complications (ie, myelodysplasia, acute myelogenous leukemia) were observed in 5 of the 77 patients followed for more than 2 years after ATG treatment. In addition, one case of non-Hodgkin's lymphoma and three solid tumors occurred in the ATG-treated patients. The survival of 56 ATG-treated patients with SAA or VSAA between the ages of 16 and 43 did not differ significantly from that of 55 adult AA patients who underwent bone marrow transplant (BMT) during the same time period (P = 0.6). However, 6-year survival rates improved from 43% for patients transplanted before 1984, to 72% for those who underwent BMT between 1984 and 1989. In contrast, there was no difference in the survival rates of patients treated with ATG during these two time periods (46% v 45%, respectively). The results suggest a superior long-term outcome for adult patients with SAA treated with BMT rather than with ATG alone, using current protocols.