Post-remission options were compared in a population of 262 relapsing and refractory acute myeloid leukemia patients achieving complete remission (CR) after the same re-induction according to etoposide - mitoxantrone - cytarabine (EMA) trials. The selection of post-remission therapy depended on trial recommendations, age, performance status, and availability of an HLA-identical sibling. One hundred and thirty patients received chemotherapy consolidation courses, 50 received autologous stem cell transplantation (SCT), and 43 underwent allogeneic bone marrow transplantation (BMT), while 39 did not receive any additional therapy. The preliminary analysis identified 3 favorable prognostic factors correlated with event-free survival (EFS): M3 subtype, previous CR duration > 1 year, and transplantation. Three year EFS was 68 vs. 23% with autologous SCT and allogeneic BMT in M3 patients and, respectively, 41 vs. 20% in non-M3 patients. Three year probabilities of treatment-related mortality were 11 and 47%, respectively. A statistical model was conceived with adjustment on prognostic factors and post-remission option. In the multivariate analysis, autologous SCT appeared significantly better than allogeneic BMT (P < 0.01) or chemotherapy (P = 0.001), while allogeneic BMT was not statistically different than chemotherapy. This indicates a high treatment-related toxicity with allogeneic BMT in patients re-induced by highly intensive chemotherapy, and therefore a tendency for a better outcome with autologous SCT as post-remission treatment in those patients.