Limiting-dilution analysis of long-term culture-initiating cells (LTCIC) is a quantitative method of estimating hematopoietic stem cell activity in clinical samples. We compared the numbers of LTCIC in bone marrow (BM), umbilical cord blood, and blood progenitor cells (obtained from patients with solid tumors at leukapheresis after mobilization with induction chemotherapy and filgrastim administration), using a two- stage long-term culture system and a limiting-dilution technique, scoring cobblestone areas of greater than 15 hematopoietic cells weekly for up to 8 weeks. Samples were obtained from 30 normal BMs, 20 human umbilical cords, and 32 leukapheresis products. Direct comparison of LTCIC in the three sources showed that the median proportions of cells generating hematopoietic foci from unfractionated mononuclear cells at 5 and 8 weeks, respectively, were 1:13,314 and 1:33,949 for BM, 1:12,506 and 1:34,546 for umbilical cord blood, and 1:10,302 and 1:12,891 for leukapheresis product. The estimated proportions of LTCIC from unfractionated mononuclear cells and CD34+ cells were similar in experiments with leukapheresis product. Leukapheresis product was superior to umbilical cord blood and cord blood to BM at 5 and 8 weeks of culture (P = .01). In two-stage long-term cultures, more colonies per flask and CD34+ cells were found in assays of leukapheresis product than in BM or umbilical cord blood cultures (P = .0005). Results obtained by this simplified limiting-dilution analysis correlated well with standard long-term cultures and can be used as a measure of the stem cell population. These data suggest that the incidence of putative stem cells in leukapheresis product and umbilical cord blood are at least comparable with that of BM.