In November 1987, the French group GOELAM initiated a randomized study comparing allogeneic bone marrow transplantation (BMT), autologous bone marrow transplantation (ABMT) and intensive consolidation chemotherapy (ICC). The induction treatment was randomized between Idarubicin plus Cytarabine and Zorubicine plus Cytarabine: 223 patients with de novo AML and aged 15-50 years are currently evaluable and 178 of them (80%) have achieved complete remission (CR) with no significant difference between both arms. Forty four patients under 40 years of age and having a HLA identical sibling were assigned to BMT and 38 were actually transplanted. Thirty of the 134 other patients did not receive the planned first course of ICC, 4 patients died during this course, and 21 were excluded before randomisation. Thus, only 64 patients have currently been randomized between the 2nd course of ICC (34 patients) and ABMT (30 patients). ABMT was prepared by the Baltimore regimen and the marrow was unpurged. With a median follow-up time of 29 months, the actuarial risk of relapse at 3 years is 29% for BMT, 38% for ABMT and 53% for ICC. The 3 year disease free survival (DFS) is 51% for BMT, 62% for ABMT and 47% for ICC. These differences are not statistically significant. When intention to treat is considered, there is no difference in the actuarial DFS between the BMT and the non BMT groups. Longer follow-up time and larger number of patients are warranted to demonstrate any significant advantage of one of these approaches.