We analysed the results of three protocols from 1990 to 2005. Protocol I (1990-1996) consisted of a 2 year VAPA regime. Protocol II (1996-2003) on 1 year daunorubicin/cytarabine alternating with etoposide/cytarabine. Protocol III (2003-2005) on six cycles MRC AML 10 modified. Patients with de novo acute myeloid leukemia 0 to 18 years were included. Demographic and clinical characteristics were analysed. Patients with >100,000 leukocytes, M4 or M5 and primary CNS disease were considered high risk. We compared remission rate, overall and event-free survival. Descriptive statistics, chi square, Kaplan-Meier and long rank tests were used. One hundred forty-five patients were included, 46 in Protocol I; 60 in II and 39 in III. There were no differences in characteristics between groups, except for more low risk patients in Protocol II (61%vs. 43% and 41%. (p = 0.05). Remission rate for Protocol I was 52%, for II 50% and for III 92% (p = 0.0001). Relapse was 18, 30 and 35, respectively (p = 0.141). Five-year event-free survival was 17.9% +/- 6.6%, 15.5% +/- 4.1% and 43.5% +/- 4.1% (s.e) (p = 0.0002). Five-year overall survival was 19.5% +/- 8%, 17.2% +/- 5.9% and 51.2% +/- 4.1% (s.e) (p = 0.0002). The results were superior in the MRC-10 derived protocol.