Bone marrow transplants for severe aplastic anemia were first performed in the 1970s. Transplant regimens, supportive care, and patient selection have changed substantially since then. Our objective was to determine the impact of these changes on transplant outcome. We studied 1,305 recipients of HLA-identical sibling transplants for aplastic anemia between 1976 and 1992, reported to the IBMTR by 179 centers. We compared survival of transplants performed in three intervals (1976 through 1980 [n = 186], 1981 through 1987 [n = 648], and 1988 through 1992 [n = 471]) using Cox proportional hazards regression. Five-year survival (+/-95% confidence interval) increased from 48% +/- 7% in the 1976-1980 cohort to 66% +/- 6% in the 1988-1992 cohort (P < .0001). Risks of graft-versus-host disease (GVHD) and interstitial pneumonia decreased over time, but the risk of graft failure did not. Higher long-term survival resulted primarily from decreased mortality in the first 3 months posttransplantation. Late mortality risks were low and changed little over the intervals studied. In multivariate analysis, changes in transplantation strategies accounted for most but not all of the improved outcome. Use of cyclosporine to prevent GVHD was the most important factor. Changes in patient selection did not seem to explain improved survival. Survival after HLA-identical sibling bone marrow transplantations for aplastic anemia has improved since 1976. Changes in GVHD prophylaxis account for much of this improvement. Other changes may also operate.