The purpose of this study was to develop a simple assay for quantitating transplantable human lymphomyeloid stem cells (competitive repopulating units [CRU]) to enable comparison among the numbers and types of progeny generated in NOD/ SCID mice by such cells from different ontologic sources. Sub-lethally irradiated NOD/SCID mice were transplanted with varying numbers of CD34+ cell-enriched suspensions of human fetal liver, cord blood, or adult marrow cells. The types and numbers of human cells present in the marrow of the mice were measured 6 to 8 weeks later using flow cytometry, in vitro progenitor assays, and secondary transplant endpoints. Frequencies of human CRU obtained by limiting dilution analysis of mice repopulated 6 to 8 weeks posttransplant were the same when the lymphoid and myeloid progeny of CRU were both detected by specific immunophenotypic endpoints as when in vitro myeloid progenitor assays were used to detect CRU myelopoietic activity. The average output per injected CRU of very primitive cells (CD34(+)CD38(-) cells, LTC-IC, and secondary CRU) was found to be highest for fetal liver CRU and progressively decreased (up to >100-fold) for ontologically older CRU. In contrast, the average output of mature cells was highest for cord blood CRU and lowest for fetal liver CRU, despite equivalent production of intermediate progenitors. Differences in the relative numbers of mature lymphoid, myeloid, and erythroid progeny produced by CRU from different ontologic sources were also seen. Finally, evidence of a transplantable human lymphoid-restricted cell present throughout ontogeny was obtained. A simpler and easier assay for enumerating transplantable human stem cells with lymphomyeloid reconstituting activity has been described, and its specificity and sensitivity validated. The use of this assay has revealed ontogeny-associated differences in a variety of functional attributes of human stem cells proliferating and differentiating in an in vivo, but xenogeneic, setting.