Since January 1989, 56 patients (31 females and 25 males) with de novo acute myelogenous leukemia have been included in the study. Their median age was 43 years (range, 15 to 60 years) with a distribution according to French-American-British morphologic subtypes as follows: six M1, 14 M2, four M3, 19 M4, nine M5, two M6, and two M7. The induction regimen (IDAC) consisted of idarubicin (12 mg/m2/d intravenously [IV] days 1 to 3) in combination with cytarabine (100 mg/m2/d continuous IV days 1 to 7). Patients achieving complete remission (CR) or partial remission received another cycle of IDAC followed by NOVE (mitoxantrone 10 mg/m2/d IV days 1 to 5 and etoposide 100 mg/m2/d IV days 1 to 5). Fifty-four patients are evaluable for response: after two cycles of IDAC, 42 patients had attained CR (78%), while 76% of these had already reached CR after the first cycle. Of the initial 11 nonresponders to IDAC, four obtained CR after NOVE. Thus, 46 of 54 patients (85%) achieved CR after sequential treatment with IDAC and NOVE. In the last 17 patients who entered CR or partial remission after the first cycle of IDAC, recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF; 3 micrograms/kg/d) was administered for 6 days starting 3 days prior to the second cycle of IDAC. For consolidation with NOVE, rhGM-CSF was given according to the same dosage schedule. After 72 hours of rhGM-CSF treatment, the white blood cell count showed a median 3.9-fold increase, without appearance of myeloblasts in the peripheral blood. During sequential chemotherapy, no significant complications (in particular, no major cardiac toxicity) were observed. Postremission, patients were either given bone marrow transplants, received late consolidation with high-dose cytarabine/mitoxantrone, or were followed up without any further treatment. Of the 46 patients evaluable for disease-free survival, 21 patients (45%) remain in CR with a 34% probability of disease-free survival at 37 months. The response-adapted treatment with IDAC/NOVE is effective and very well tolerated. To define the therapeutic impact of rhGM-CSF, a randomized trial will be required.