The genetic architecture of cancer has been delineated through advances in high-throughput next-generation sequencing, where the sequential acquisition of recurrent driver mutations initially targeted towards normal cells ultimately leads to malignant transformation. Myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) are hematologic malignancies frequently initiated by mutations in the normal hematopoietic stem cell compartment leading to the establishment of leukemic stem cells. Although the genetic characterization of MDS and AML has led to identification of new therapeutic targets and development of new promising therapeutic strategies, disease progression, relapse, and treatment-related mortality remain a major challenge in MDS and AML. The selective persistence of rare leukemic stem cells following therapy-induced remission implies unique resistance mechanisms of leukemic stem cells towards conventional therapeutic strategies and that leukemic stem cells represent the cellular origin of relapse. Therefore, targeted surveillance of leukemic stem cells following therapy should, in the future, allow better prediction of relapse and disease progression, but is currently challenged by our restricted ability to distinguish leukemic stem cells from other leukemic cells and residual normal cells. To advance current and new clinical strategies for the treatment of MDS and AML, there is a need to improve our understanding and characterization of MDS and AML stem cells at the cellular, molecular, and genetic levels. Such work has already led to the identification of promising new candidate leukemic stem cell molecular targets that can now be exploited in preclinical and clinical therapeutic strategies, towards more efficient and specific elimination of leukemic stem cells.