Agnogenic myeloid metaplasia (AMM) is a clonal stem cell disorder that leads to ineffective hematopoiesis, bone marrow fibrosis, and extramedullary hematopoiesis. The molecular mechanisms underlying the development of this syndrome are currently unknown. Therefore, the aim of this study was to characterize aberrant gene expression in CD34+ hematopoietic stem cells from patients with AMM. We used oligonucleotide microarrays to analyze gene expression profiles in CD34+ hematopoietic stem cells from patients with AMM compared with expression in CD34+ cells from healthy individuals. We identified 95 highly differentially expressed genes (48 upregulated and 47 down-regulated) that are potentially involved in regulating abnormal hematopoietic proliferation and differentiation and confirmed many of them by quantitative polymerase chain reaction. Using class membership prediction analysis, we identified 75 genes whose expression profiles can accurately differentiate AMM samples from the controls. Using these 75 genes, we were able to discriminate patients with AMM from the controls by hierarchical clustering (Spearman's confidence correlation). The predictive power of these genes was verified by applying the algorithm to an unknown test set containing expression data from eight additional CD34+ samples (four AMM, four control). Our results indicate that a subset of genes may be used to differentiate patients with AMM from healthy individuals. Furthermore, we identify 95 genes whose aberrant expression may be involved in AMM.