The main difference between the two cooperative studies on therapy of childhood acute myelogenous leukemia AML-BFM-78 and AML-BFM-83 was the addition of an 8-day ADE (cytosine arabinoside, daunorubicin, etoposide) induction treatment in the second study. Due to this intensification, the relapse rate, but not the rate of induction failures, was reduced. The probability of a 6-year event-free survival increased from 38%, SD 4%, in study AML-BFM-78 to 49%, SD 4%, in study AML-BFM-83, P = .08. The improvement of the 6-year event-free interval (EFI) was significant in the second study (61%, SD 4%, versus 47%, SD 5%, P less than .05); it was restricted to the FAB types M1 through M4 (EFI: 67%, SD 5%, versus 45%, SD 5%, P less than .01). The difference in EFI seen in FAB M5 was not statistically significant (EFI: 40%, SD 10%, versus 63%, SD 11%, NS). According to the results of the second study, two different risk groups (low and high) could be identified by combinations of predominantly pretherapeutic parameters. The low risk group, comprising 37% of the patients who achieved complete remission, included the FAB types with granulocytic differentiation and specific additional features: FAB M1 with Auer rods, FAB M2 with white blood cell count of less than 20,000/microL, FAB M3 all patients, and FAB M4 with eosinophilia. The 6-year Kaplan-Meier estimation of EFI is 91%, SD 4%, compared with 42%, SD 6% in the high risk group. In future studies based on the AML-BFM-83 treatment, bone marrow transplantation in first remission should be mandatory only for children of the high risk group.