In order to test the hypothesis that the telomere length in human hematopoietic cells correlates with their proliferative potential, we analyzed the telomere length in highly purified subpopulations of bone marrow cells. Cells were sorted on the basis of CD34 and CD38 cell surface markers, and two samples were additionally sorted on the basis of Hoechst 33342 dye efflux allowing isolation of side population (SP) cells. The telomere length in limiting numbers of sorted cells was analyzed using a newly developed fluorescence in situ hybridization (flow-FISH) method in which hybridization of telomere probe in cells of interest is measured relative to control cells in the same tube. In all seven bone marrow samples analyzed, the telomere length in CD34(+)CD38(-) cells was longer than in CD34(+)CD38(+) cells from the same donor (p < 0.02). Results with sorted SP cells were less clear: the telomere fluorescence in these cells was very heterogeneous, and a reproducible difference in telomere length relative to CD34(+)CD38(-) cells could not be observed. We conclude that the telomere length in subpopulations of hematopoietic cells does appear to be correlated with the known proliferative potential of such cells and that further characterization of cells on the basis of telomere length is warranted for enrichment of very rare precursors of hematopoietic and other tissues.