We have developed a limiting-dilution assay of long-term repopulating hematopoietic stem cells in the mouse using a miniturized stroma- dependent bone marrow culture assay in vitro. The cells were overlaid on irradiated stromal layers in microtiter wells in a range of concentrations, and frequencies of cobblestone area-forming cells (CAFC) were calculated by employing Poisson statistics. The production of secondary granulocyte/macrophage colony-forming units (CFU-G/M) in the adherent layer of individual wells was correlated with the presence of such cobblestone areas. CAFC frequencies were determined in bone marrow cell suspensions that were either enriched for marrow repopulating ability (MRA) in vivo, while depleted for spleen colony- forming units (CFU-S), or vice versa. The separation of bone marrow cells (BMC) was either based on centrifugal elutriation, or monoclonal antibody-mediated magnetic depletion of cells carrying cell surface differentiation antigens, and subsequent sorting on the basis of light scatter and rhodamine-123 retention as a measure of mitochondrial activity. In addition, 5-fluorouracil-resistant BMC were studied. Our investigations show that a time-dependent cobblestone area formation exists that reflects the turnover time and primitiveness of CAFC. The frequency of precursors forming cobblestone areas on day 28 after overlay is proposed to be a measure for MRA, whereas the day-7 CAFC frequency closely corresponds with day-12 CFU-S numbers in the suspensions tested.