In a up-front randomized study, 939 adult patients up to the age of 60 years received a double induction therapy. One course of MAV (mitoxantrone 10 mg/m2 days 4–8, cytarabine 100 mg/m2 continuous infusion days 1–8, etoposide 100 m g/m2 days 4–8) was followed by one cycle of MAMAC (cytarabine 1 g/m2, every 12h days 1–5; amsacrine 100 mg/m2 days 1–5). Patients with intermediate risk cytogenetic (IRCG) and a HLA matched sibling received an allogeneic transplantation, those with poor risk cytogenetic (PRCG) were intended to be transplanted from a sibling or unrelated donor. All AML patients without an available donor received the randomly assigned first postremission therapy (PRT) mitoxantrone combined with intermediate-dose cytarabine (I-MAC; total dose 12 g/m2) or high-dose cytarabine (H-MAC; total dose 36 g/m2). As second PRT, patients with t(8;21) received an additional cycle of chemotherapy. An autologous transplantation was scheduled for IRCG and PRCG without an allogeneic donor. The CR rate was 88% for patients with t(8;21), with IRCG 71%, and 50% with PRCG. The 5-year-survival was 21% (95% CI: 16–27%) in the PRCG, 40% (95% CI: 36–45%) in the IRCG and 74% (95% CI: 60–88%) in the t(8;21) group. No difference was observed between the I-MAC and the H-MAC group. In a multivariate analysis, a significant (p<.01 for each parameter) better overall survival was observed in patients under the age of 37 years, blast count <10% at day 15, high myeloperoxidase positivity, low CD34 expression, WBC <15*10^9/L, thrombocytes >50*10^9/L, and IRCG compared to PRCG. The relapse incidence was higher in patients without an allogeneic donor, a Flt3 mutant/wildtype ratio > 0.8 or PRCG. A risk score build out of the sum of the individual hazard ratios (SHR) was able to discriminate two groups for the IRCG with a marked difference in the 5-year-survival (low SHR: 55% 95% CI: 48–62%; high SHR: 33% 95% CI: 28–38%) was well as for the PRCG group (low SHR: 44% 95% CI: 32–56%; high SHR: 13% 95% CI: 7–18%). The risk score identified in this large patient cohort may allow individual tailoring of therapeutic interventions in future AML trials.