We analyzed the effects of T-cell depletion on the outcome of HLA-identical sibling bone marrow transplants for leukemia by comparing 731 T-cell-depleted transplants with 2,480 non-T-cell-depleted transplants. T-cell depletion decreased acute graft-versus-host disease (GVHD) (relative risk [RR] 0.45; P less than .0001) and chronic (GVHD) (RR 0.56; P less than .0001). However, it increased graft failure (RR 9.29; P less than .0001). Leukemia relapse also was increased. In first remission acute leukemia or chronic phase chronic myelogenous leukemia, leukemia relapse was 2.75 times more likely after T-cell-depleted transplants (P less than .0001). T-cell depletion increased the risk of treatment failure (RR 1.35; P less than .0003) and decreased leukemia-free survival. We also studied controllable variables associated with outcome of T-cell-depleted transplants. The unique findings were that among recipients of T-cell-depleted transplants for early leukemia, radiation doses greater than or equal to 11 Gy (RR 0.54; P less than .01), dose rates greater than 14 cGy/min (RR 0.56; P less than .002), and additional posttransplant immune suppression with cyclosporine alone (RR 0.53; P less than .0006) or cyclosporine plus methotrexate (RR 0.36; P less than .01) were associated with fewer treatment failures. Use of monoclonal antibodies rather than physical techniques for T-cell depletion (RR 2.01; P less than .03) and fractionated radiation (RR 1.69; P less than .05) were associated with increased treatment failure and lower leukemia-free survival. These data may be useful in designing strategies to improve results of T-cell-depleted transplants.