Intensive, myelosuppressive therapy is necessary to maximize outcomes for patients with acute myeloid leukemia (AML). A comparison was made of 3 aggressive postremission approaches for children and adolescents with AML in a randomized trial, CCG-2891. A total of 652 children and adolescents with AML who achieved remission on 2 induction regimens using identical drugs and doses (standard and intensive timing) were eligible for allocation to allogeneic bone marrow transplantation (BMT) based on matched related donor status (n = 181) or randomization to autologous BMT (n = 177) or to aggressive high-dose cytarabine-based chemotherapy (n = 179). Only 115 patients (18%) refused to participate in the postremission phase of this study. Overall compliance with the 3 allocated regimens was 90%. At 8 years actuarial, 54% +/- 4% (95% confidence interval) of all remission patients remain alive. Survival by assigned regimen ("intent to treat") is as follows: allogeneic BMT, 60% +/- 9%; autologous BMT, 48% +/- 8%; and chemotherapy, 53% +/- 8%. Survival in the allogeneic BMT group is significantly superior to autologous BMT (P =.002) and chemotherapy (P =.05); differences between chemotherapy and autologous BMT are not significant (P =.21). No potential confounding factors affected results. Patients receiving intensive-timing induction therapy had superior long-term survival irrespective of postremission regimen received (allogeneic BMT, 70% +/- 9%; autologous BMT, 54% +/- 9%; chemotherapy, 57% +/- 10%). Allogeneic BMT remains the treatment of choice for children and adolescents with AML in remission, when a matched related donor is available. For all others, there is no advantage to autologous BMT; hence, aggressive nonablative chemotherapy should be used.