Two novel preparatory regimens for conditioning of patients with leukemia for allogeneic bone marrow transplantation (BMT) from histocompatible sibling donors have been tested in a phase III trial under the auspices of the Southwest Oncology Group (SWOG 8612). These two regimens consisted either of fractionated total body irradiation and etoposide (FTBI/VP-16) or high-dose busulfan with cyclophosphamide (BU/CY). Only patients who had failed prior conventional management at least once were study eligible, ie, no patients with acute leukemia in first remission (CR) or in first chronic phase (CP) of chronic myelogenous leukemia (CML) participated. Patients were stratified according to the following risk criteria: "good-risk" patients were those who were in second CR of their acute leukemia or in accelerated phase (AP) of CML; "poor-risk" patients had further advanced stages of leukemia. During a 52-month period, 131 patients were registered of whom 122 (93%) were study eligible. Sixty-one eligible patients were randomized to the FTBI/VP-16 arm and 61 to the BU/CY regimen. Of these 122 patients, 114 (93%) proceeded to BMT according to protocol. Posttransplant immunosuppression to prevent graft-versus-host disease (GVHD) consisted of cyclosporine and prednisone (CSA/PSE). Neither overall survival nor disease-free survival (DFS) differed significantly between the two treatment groups (P = .89 and .69, respectively). Estimated DFS for "good-risk" patients who had been prepared with the FTBI/VP-16 regimen was 55% +/- 11%, as compared with patients treated with BU/CY whose DFS figure was 34% +/- 10% (P = .30). For "poor-risk" candidates, the DFS rates at 24 months were 17% +/- 6% (for FTBI/VP-16) and 24% +/- 8% (for BU/CY), respectively (P = .81). These figures do not differ significantly, especially in view of the fact that the "good-risk" patients prepared with the FTBI/VP-16 regimen were younger than those treated with BU/CY. Both regimens were well tolerated with no regimen-related deaths encountered during the 6-week period after BMT. This study also confirmed the efficacy of the CSA/PSE combination in the prevention of GVHD with 23 of 113 (20%) of BMT recipients developing moderate to severe acute GVHD. The leading cause for treatment failure was leukemic relapse (45 of the 114 BMT recipients suffered a recurrence of their leukemia), whereas 38 patients died without evidence of relapse. Thirty-one patients are alive and in continued CR after marrow transplantation; four are alive in relapse.(ABSTRACT TRUNCATED AT 400 WORDS)