Analysis of molecular mechanisms associated with stem cell commitment and differentiation requires an in vitro assay that identifies the most primitive hematopoietic stem cells in human bone marrow. Such primitive stem cells usually do not form colonies in short-term semisolid assays and are best identified by their ability to initiate sustained hematopoiesis when they are cocultured with competent stromal cells. In this study, we investigated whether a murine marrow stromal cell line (MS-5) that supports colony-forming unit-spleen (CFU-S) maintenance would permit, both in short-term colony assays and long-term cultures, the development of primitive human stem cells sorted on the basis of their high expression of CD34 and lack of expression of CD38 antigen. In short-term colony assays, this population included almost exclusively primitive progenitor cells. MS-5 cells synergized with any combination of interleukin-3, Steel factor, granulocyte colony- stimulating factor, agar-leukocyte conditioned medium, and erythropoietin and increased at least twofold both the cloning efficiency of CD34++/CD38- cells and the size of the colonies. Furthermore, MS-5 cells triggered the development of multipotent blast cell progenitors with a high proliferative potential, which in these conditions represented 1% to 2% of CD34++/CD38- cells. When MS-5 cells were substituted by human stromal cells or when growth factor combinations were used in the absence of stromal cells, much lower numbers of CFU-blast were detected. This selective action of MS-5 on early progenitors was also observed when MS-5 cells were used as feeders in long-term cultures of CD34++/CD38- cells. Murine cells promoted the expansion of high proliferative potential primitive progenitor cells up to 3 months, although they did not support their differentiation in mature clonogenic progenitors or terminally differentiated cells. Sustained hematopoiesis in these longterm cultures was accounted for by 2% to 5% of initial CD34++/CD38- cells as estimated by limiting dilution experiments. Mechanisms by which murine stromal cells act specifically on human primitive stem cells are unclear, but from our data this effect is unlikely to be explained solely by known species cross-reactive growth factors. Further manipulation of this long-term coculture system should prove useful in identifying stromal molecules regulating commitment and differentiation of early human progenitor cells.