Cytogenetics is still being considered the most powerful single prognostic factor, which is useful to determine the types of post-remission therapy in AML, though various molecular markers are available for predicting the prognosis of AML patients. Most phase III studies have failed to demonstrate a clear advantage of allografting over chemotherapy in terms of overall survival because of significant risk of transplant-related mortality. Optimal post-remission therapies in terms of frequencies (number of treatment) or intensities are not decided yet. In this study, since 2000, we investigated that outcomes of post-remission therapies(high-dose cytarabine (HDAC) vs autologous stem cell transplantation (AutoSCT) vs allogeneic stem cell transplantation from sibling or unrelated donors (AlloSCT)) based on cytogenetic risk (GPG, Good prognosis group; IPG, Intermediate prognosis group; PPG, Poor prognosis group by MRC definition) on the AML patients who achieved complete remission after induction chemotherapy. The aims of this prospective intention to treat analysis was to compare the CR, recovery kinetics, DFS and OS in the different prognostic groups. Three plus seven (idarubicin 12mg/m2, D1–D3; cytarabine 100mg/m2, D1–D7) were given to de novo AML, secondary AML and therapy-related AML. Then, HDAC or AutoSCT was given after intermediate dose (8gm/m2) of cytarabine to the patients with GPG. Three times of post-remission therapy including HDAC, or AutoSCT followed by two times of post-remission therapy were given to IPG or PPG. If HLA-identical sibling was available, then AlloSCT underwent after 1st post-remission therapy. Since January, 2000, 506 patients(18 centers) were enrolled up to December, 2007. Among them, 92.3% was de novo AML, and GPG, IPG and PPG were, 23.1%, 62.1% and 14.8% respectively. Over all complete remission rate after 1st induction was 79.0% and CR rate in GPG, IPG, PPG were 92.0%, 81.0% and 43.9% respectively(P<0.001) in 476 patients who were eligible to this study. In Good Prognosis Group (GPG), survivals were not different between different treatment groups (5 year LFS: HDAC 34.2%, AutoSCT 63.5%, AlloSCT 54.8%, p=0.270; 5 year OS: HDAC 54.5%, AutoSCT 62.5%, AlloSCT 53.3%, p=0.676). However, beneficial effect of AlloSCT in post-remission therapy therapy was observed by multivariate analysis in terms of LFS compared to HDAC (HR of relapse for HDAC 3.198 compared to AlloSCT, p=0.045). Outcomes of HDAC group were inferior in GPG in terms of OS and LFS compared to other studies. This results may be due to low cumulative dose of Ara C, because patients of HDAC group in GPG treated just 1 cycle of IDAC before HDAC therapy. In addition, in our cohort, majority (80%) of GPG have t(8;21), which are known as having inferior survival results, compared to inv(16) group. In Intermediate Prognosis Group (IPG), survivals were not different among different types of treatment (5 year LFS: HDAC 31.1%, AutoSCT 42.4%, AlloSCT 55.0%, p=0.131; 5 year OS: HDAC 39.2%, AutoSCT 42.5%, AlloSCT 46.5%, p=0.491). AlloSCT group showed a trend of being superior to other therapeutic modalities in terms of LFS (p=0.07). AutoSCT group showed a trend of being superior to other therapeutic modalities in OS by multivariate analysis (HR of death for AutoSCT 0.539 compared to AlloSCT, p=0.085). In Poor Prognosis Group (PPG), though data showed slightly beneficial effect of AlloSCT in AML therapy, however, there were no significant statistical differences on OS/LFS in 3 types of consolidation therapy modalities (4 year LFS: HDAC 48.3%, AutoSCT 0%, AlloSCT 39.1%, p=0.379; 4 year OS: HDAC 21.4%, AutoSCT 33.3%, AlloSCT 56.1%, p=0.638). Based on this trial, Allo- or Auto-SCT over HDAC may have beneficial effects in some subgroup with high risk and young age, among the patients with good and intermediate cytogenetic risk. In GPG, “sufficient cumulative dose” of Ara C seems to be necessary to have a good outcome. However, GPG seems to be heterogenous group in terms of biology having poor prognosis when one has additional CG abnormalities on top of t(8;21) or inv(16), which ones need to investigate further. While finding more effective anti-AML molecules/monoclonal Ab's are necessary, good therapeutic rationales in terms of choosing AlloSCT vs AutoSCT vs HDAC should be established. Same time, identifying for better cellular and molecular prognostic factors over cytogenetics are still relevant for designing “effective therapies, but minimal toxicities”.