The failure of complete remissions to reliably translate into cures in acute myeloid leukemia (AML) can be explained by the leukemia stem cell (LSC) paradigm, which hypothesizes that rare leukemia cells with stem cell features, including self-renewal capacity and drug resistance, are primarily responsible for both disease maintenance and relapses. Traditionally, the ability to generate AML in immunocompromised mice wereÂ how these so-called LSCs were identified. Only those rare AML cells characterized by a hematopoietic stem cell (HSC) CD34+CD38− phenotype were believed capable of generating leukemia in immunocompromised mice, but more recently, significant heterogeneity in the phenotypes of engrafting AML cells has been demonstrated. Moreover, AML cells that engraft immunocompromised mice do not necessarily represent either the founder clone or those cells responsible for relapse. A recent study found that the most immature phenotype present in an AML was heterogeneous, but correlated with genetically defined risk groups and outcomes. Patients with AML cells expressing a primitive HSC phenotype (CD34+CD38− with high aldehyde dehydrogenase activity) manifested significantly lower complete remission rates, as well as poorer event-free and overall survivals. AMLs in which the most primitive cells displayed more mature phenotypes were associated with better outcomes. The strong clinical correlations suggest that the most immature phenotype detectable within a patient’s AML might serve as a biomarker for “clinically relevant” LSCs. The minimal residual disease state during first remission may be the optimal setting to study novel LSC-targeted therapies, since they may have limited activity against the bulk leukemia and will be utilized at lowest tumor burden as well as least tumor heterogeneity.