Between 1990 and 1996, we conducted a randomized trial in adults with newly diagnosed acute myeloid leukemia (AML) in order to compare relapse-free interval (RFI) after double induction (arm B), timed-sequential induction (arm C), or control "3 + 7" induction (arm A). Patients achieving complete remission (CR) after induction +/- salvage received the same consolidation chemotherapy, which included a dosage stratification according to patient's age (younger or older than 50 years). This long-term analysis was performed in 592 patients (arm A/B/C, 197/198/197 patients). Overall CR rate was 76% without differences between the 3 arms, even if a salvage course was less frequently needed in arm B. Treatment-related mortality, either during the induction or the postremission phase, was not significantly higher in arms B and C than in arm A. Among the 449 CR patients, 250 relapsed (arm A/B/C, 90/87/73 patients) without significant differences in RFI in arms B and C versus arm A (P = .39 and .15, by the Gray test). However, when analyzing the 345 patients younger than 50, RFI was significantly improved in younger patients receiving timed-sequential induction (P = .038 by the Gray test), while not in those receiving double induction. Event-free survival and overall survival were similar in the 3 randomization arms.