Using a genetic randomization through a 'donor' vs 'no donor' comparison, the aim of this analysis was to assess the real benefit of reduced intensity conditioning allogeneic stem cell transplantation (RIC-allo-SCT) among 95 adult high-risk acute myeloid leukemia (AML) patients. In an 'intention-to-treat' analysis, leukemia-free survival (LFS) was significantly higher in the 'donor' group as compared to the 'no donor' group (P=0.01; 54 vs 30% at 4 years). The latter held true when restricting the analysis to the 25 patients who could actually receive the RIC-allo-SCT (P=0.001). Overall transplant-related mortality in the 'transplant' group was 12%, with overall survival (OS) being significantly higher in the 'transplant' group as compared to the 'no transplant' group (P=0.01). Also, in the 'intention-to-treat' analysis, OS was significantly higher in the 'donor' group as compared to the 'no donor' group (P=0.04). In the multivariate analysis, actual performance of RIC-allo-SCT (P=0.001; RR=4.0; 95% CI, 1.7-9.6) was the strongest factor significantly predictive of an improved LFS. We conclude that if a matched related donor is identified, RIC-allo-SCT should be proposed for AML patients not eligible for standard myeloablative allo-SCT.