The cellular and molecular basis for intra-tumoral heterogeneity is poorly understood. Tumor cells can be genetically diverse exhibiting intra-tumoral functional heterogeneity. Often proposed as mutually exclusive, cancer stem cell (CSC) models postulate that tumors are cellular hierarchies created due to epigenetic programs that are sustained by CSC. I will focus on three lines of evidence showing these models are highly integrated. Gene signatures specific to AML LSC have revealed a common stemness program that is highly predictive of patient survival in clinical databases of >1000 samples. Thus, determinants of stemness influence clinical outcome of AML across a spectrum of mutations indicating that many genetic abnormalities coalesce around stem cell properties. Secondly, combined genetic and functional studies of the LSC from either B-ALL or AML point to commonalities between clonal evolution and CSC models of cancer. LSC originate genetically diverse subclones that are related through a complex branching evolutionary process, specific mutations influence LSC functions. Thus the clonal evolution models are highly relevant in cancer but need to be extended to adopt the concept that CSC are subject to clonal evolutionary forces. Finally, the combined genetic and functional analysis of both AML blast cells as well as the non-leukemic hematopoietic cells isolated from the same patient blood samples is revealing fundamental insights into the cell of origin, nature and biological consequences of initiating lesions and order of subsequent mutations; concepts that demonstrate how highly integrated the CSC and genetic evolution models must be. HSC, progenitor and mature cell fractions contained recurrent mutations but without other mutations present in AML blasts. These pre-leukemic-HSC were clonally expanded and survived chemotherapy, remaining present in remission and relapse samples. Detailed studies on 11 paired Dx/Rel samples followed by detailed genetic tracking in sorted blasts and progenitor subpopulations as well as in LSC derived xenografts allowed us to trace the origins of Rel. In all cases, the Rel clone was pre-existing in the Dx blood sample and often originating in rare LSC subclones distinct from the dominant blasts. Thus our findings indicate the future therapeutic strategies must account for and target rare genetic subclones, as well as the functionally important LSC and preL-HSC that are solely capable of clonal propagation.