In order to identify acute myeloid leukemia (AML) CD34 þ -specific gene expression profiles, mononuclear cells from AML patients (n ¼ 46) were sorted into CD34 þ and CD34 À subfractions, and genome-wide expression analysis was performed using Illumina BeadChip Arrays. AML CD34 þ and CD34 À gene expression was compared with a large group of normal CD34 þ bone marrow (BM) cells (n ¼ 31). Unsupervised hierarchical clustering analysis showed that CD34 þ AML samples belonged to a distinct cluster compared with normal BM and that in 61% of the cases the AML CD34 þ transcriptome did not cluster together with the paired CD34 À transcriptome. The top 50 of AML CD34 þ -specific genes was selected by comparing the AML CD34 þ transcriptome with the AML CD34 À and CD34 þ normal BM transcriptomes. Interestingly, for three of these genes, that is, ankyrin repeat domain 28 (ANKRD28), guanine nucleotide binding protein, alpha 15 (GNA15) and UDP-glucose pyrophosphorylase 2 (UGP2), a high transcript level was associated with a significant poorer overall survival (OS) in two independent cohorts (n ¼ 163 and n ¼ 218) of normal karyotype AML. Importantly, the prognostic value of the continuous transcript levels of ANKRD28 (OS hazard ratio (HR): 1.32, P ¼ 0.008), GNA15 (OS HR: 1.22, P ¼ 0.033) and UGP2 (OS HR: 1.86, P ¼ 0.009) was shown to be independent from the well-known risk factors FLT3-ITD, NPM1c þ and CEBPA mutation status.