To compare the antileukemic efficacy of idarubicin and mitoxantrone in elderly patients with acute myeloid leukemia (AML) and to evaluate the feasibility of autologous transplantation using PBSC after consolidation in those with a good performance status, 160 patients (median age 69 years), with AML at diagnosis, 118 of them with de novo AML and 42 with AML secondary to myelodysplastic syndrome or toxic exposure (sAML), received induction treatment with idarubicin, 8 mg/m2/day or mitoxantrone, 7 mg/m2/day, on days 1, 3, and 5, both combined with VP-16, 100 mg/m2/day on days 1 to 3 and cytarabine (araC), 100 mg/m2/day, on days 1 to 7. G-CSF, 5 microg/kg/day, was administered after chemotherapy in patients aged more than 70 years. Patients in complete remission (CR) received one course of consolidation using the same schedule as for induction except the araC administration was shortened to 5 days. Some patients younger than 70 years were then scheduled for autologous stem cell harvest on days 5 to 7 of G-CSF, 5 microg/kg/day, initiated after hematopoietic recovery from consolidation. Autologous transplantation was performed following an additional chemotherapy conditioning. Ninety-five patients (59%) achieved CR, without significant difference between the idarubicin (56% CR) and mitoxantrone (63% CR) group. There was also no significant difference in CR rate between de novo AML (63%) and secondary AML (55%) (P = 0.12). Patients aged < 70 years had 67% CR, while patients aged > or = 70 years had 49% (P = 0.02). There was no significant difference in the duration of aplasia between the two arms. Median time to neutrophil recovery was 22 days in patients who received G-CSF following induction and 27 days in patients who did not (P = 0.006). Severe extrahematologic toxicities of induction did not differ between the two arms and included sepsis (39%), diarrhea (13%), hyperbilirubinemia (8%), hemorrhage (6%) and vomiting (6%). Overall, 14 patients (9%), died from toxicity of induction. First consolidation was administered in 74 patients of whom seven (9%) died from toxicity. Nineteen patients have received transplantation. Median time to recovery of neutrophils > 0.5 x 10(9)/l was 13 days and of platelets > 50 x 10(9)/l 43 days following consolidation. There were two toxic deaths. Median disease-free survival and survival from time of achieving CR of non transplanted patients are 6 and 7 months respectively without difference between the two arms. Fourteen transplanted patients relapsed at a median of 5 months post-transplant. We conclude that this regimen is well tolerated and has a good efficacy to induce CR, without a significant difference in efficacy and toxicity between idarubicin and mitoxantrone. Intensive postinduction, including transplantation, is feasible; however, this procedure did not seem to prevent early relapse in the majority of patients. Neither the high rate of CR nor consolidation nor transplant procedure in a selected group of patients did translate into improved DFS and/or survival.