OBJECTIVE To report the outcome of autologous bone marrow transplantation for patients with acute myeloid leukemia (AML) in first or greater complete remission (CR) treated by autologous bone marrow transplantation using two different preparatory regimens. METHODS Between September 1986 and August 1993, 75 patients with AML ranging in age from 6 months to 58 years underwent autologous bone marrow transplantation using previously harvested and frozen unpurged (n = 6) or 4-hydroperoxycyclophosphamide purged marrows (n = 69). Patients were in first CR (n = 44) or beyond first CR (n = 31). The preparative regimen consisted of 120 mg/kg of cyclophosphamide (CY) and 1320 cGy total body irradiation (TBI) in eight fractions over 4 days (CY/TBI) in 29 patients; and 16 mg/kg of Busulfan (BU) and 200 mg/kg of CY (BU/CY) in 46 patients. Thirty-five of these 75 patients (18 CY/TBI and 17 BU/CY) were part of a randomized trial comparing the two preparative regimens. RESULTS At 2 years, overall survival and disease-free survival (DFS) were 49% [95% confidence interval (C.I.) 37-61%] and 43% (95% C.I. 32-55%), respectively. Patients in first CR had a significantly better outcome than patients beyond first CR with an estimated 2-year DFS of 59% (95% C.I. 44-74%) vs. 21% (95% C.I. 5-36%, log-rank p = 0.0001), respectively. For patients conditioned with CY/TBI, the estimated 2-year DFS was 52% compared to 39% for BU/CY (log-rank p = 0.35). Estimated 2-year relapse rates were 44% vs. 56% (log-rank p = 0.40), respectively. For patients in first CR, no differences in DFS were observed between the two regimens (2-year estimates 69% vs. 55% log-rank p = 0.52). Patients beyond first CR had a significantly improved DFS with the CY/TBI regimen (2-year estimates of 38% vs. 7%, log-rank p = 0.04). No differences were found between the two regimens in terms of time to WBC engraftment, absolute neutrophil count of > 500, incidence of bacteremias, or median time to hospital discharge. Interstitial pneumonitis developed in two patients (one BU/CY, one CY/TBI) and venoocclusive disease developed in seven BU/CY patients (Fishers exact test p = 0.04). CONCLUSIONS For patients beyond first CR, the CY/TBI regiment provided a better outcome, with a significantly better disease-free survival and less venoocclusive disease. For patients in first CR, no significant difference between the two regimens was found. The high relapse rate, especially for patients with advanced disease, emphasizes the need for early transplantation and for new strategies to improve outcome.