From December 1987 to June 1992, 251 patients aged 50-65 with de novo acute myelogenous leukaemia (AML) were recruited to a multi-institutional randomized clinical trial. Induction therapy consisted of Ara-C (200 mg/ m2, continuous infusion, days 1-7) with either zorubicin (ZRB) (200 mg/m2, i.v., days 1-4) or idarubicin (IDR) (8 mg/ m2, i.v., days 1-5). Consolidation therapy consisted of a single course of intensive chemotherapy with high-dose Ara-C (3 g/m2, 3 h infusion, q 12 h, days 1-4) and m-Amsa (100 mg/m2/d, i.v., days 5-7). The complete remission (CR) rate was (73%) with Ara-C/ IDR versus (60%) with Ara-C/ZRB (P = 0.033). In multivariate analysis, factors found to be significant in predicting CR were normal karyotype and treatment with IDR. With a median follow-up of 73 months, the median disease-free survival (DFS) duration of all CR patients and the probability of CR at 6 years were 17 months and 29%. In multivariate analysis the only factor associated with an increased DFS duration was a normal karyotype. The median event-free survival (EFS) duration for all evaluable patients and the median overall survival duration for all eligible patients were respectively 7 and 12 months without any difference between induction arms. The study shows that in patients aged 50-65 idarabicin is more effective than zorubicin for remission induction. However, the type of anthracycline did not influence overall survival duration. Using a unique consolidation course, we observed a prolonged DFS which compares favourably with results obtained with more prolonged consolidation therapy or maintenance treatment.