We enriched bone marrow cells from 10 normal individuals for primitive hematopoietic progenitors using a two-step technique, and examined resultant primitive progenitors for their in vitro longterm repopulating capacity and their ability to adhere to irradiated stroma. Immunomagnetic depletion of mature myeloid and lymphoid progenitors resulted in a lineage-negative (Lin-) cell population. Subsequent dual-color fluorescence activated sorting of cells with low forward and vertical light scatter properties, expressing CD34 antigen (34+) and either bearing (DR+) or lacking (DR−) the HLA-DR antigen, resulted in the selection of Lin− 34+DR+ and Lin− 34+DR− cell populations. When the Lin − 34+DR+ cell fraction was cultured in a shortterm methylcellulose assay, we demonstrated a 61-fold enrichment for colony forming cells (CFC) compared with undepleted bone marrow mononuclear cells. In contrast to the Lin − 34+DR+ cells, direct culture of Lin− 34+DR− cells in short-term methylcellulose generated significantly less CFC (p ≤ 0.001). We then compared the capacity of Lin − 34+DR+ and Lin − 34+DR − cells to generate sustained hematopoiesis when plated in long-term bone marrow culture (LTBMC). When LTBMC were initiated with plated Lin −34+DR− cells, we recovered high numbers of CFC during the first week, but observed a rapid decline in the number of harvested CFC over the following weeks. No CFC could be recovered after week 7. In contrast, LTBMC initiated with plated Lin−34+DR− cells yielded significantly greater numbers of CFC than LTBMC initiated with plated Lin − 34+DR+ cells (p ≤ 0.001), and this was sustained for at least 12 wk of culture. The Lin − 34+DR+ population was only 6.6-fold enriched for primitive progenitors capable of initiating and sustaining hematopoiesis in LTBMC when compared with undepleted bone marrow mononuclear cells, while the Lin −34+DR− ‘population was 424-fold enriched for such primitive progenitors (p ≤ 0.001). Finally, we examined the capacity of both Lin − 34+DR+ and Lin −34+DR− populations to adhere to irradiated allogeneic stroma. We used a previously described "panning method" in which cells are plated onto stroma for 2 h, the nonadherent cells removed by extensive washing, and the adherent fraction maintained under conditions favoring LTBMC growth. When stroma was panned with Lin − 34+DR+ cells, 79 ± 10% of the cells were recovered in the panning effluent. In contrast, when stroma was panned with Lin − 34+DR− cells, significantly fewer (37 ± 7%) (p ≤ 0.001) cells were recovered in the panning effluent. Unlike LTBMC initiated with plated Lin −34+DR− cells, virtually no CFC were recovered from LTBMC initiated with panned Lin − 34+DR+ cells. In contrast, LTBMC initiated with either plated or panned Lin−34+DR− cells generated high numbers of CFC for a minimum of 12 wk. These studies present the first evidence that further purification of 34+/DR- cells using an additional immunomagnetic depletion of committed myeloid and lymphoid progenitors results in a Lin −34+DR− population that is significantly enriched (424- fold) for primitive progenitors capable ofinitiating and sustaining growth of committed myeloid progenitors in LTBMC for at least 12 wk. These studies also provide the first evidence that primitive progenitors capable of adhering avidly to irradiated bone marrow-derived stroma when panned for 2h are present exclusively in the Lin − 34+DR− population. In contrast, Lin −34+DR − cells, which are committed clonogenic precursors, do not exhibit the ability to adhere to irradiated stroma. Further study of these cell populations will allow detailed analysis of interactions between primitive hematopoiesic stem cells and the bone marrow microenvironment.