Replating experiments have shown that the self-renewal of pluripotent hemopoietic stem cells can be studied in vitro by clonal analysis techniques. The number of daughter stem cells detectable in individual primary clones produced in vitro varies markedly from one clone to another. These findings are consistent with a general model of stem cell differentiation in which the choice to self-replicate or not is ultimately determined at the single-cell level by a mechanism involving a random-event component that is intrinsic to the stem cell itself. Hemopoietic stem cells were identified by their ability to generate macroscopic-sized colonies having a visible erythroid component (i.e., gross red color) in standard methylcellulose assays containing medium conditioned by pokeweed mitogen-treated spleen cells and erythropoietin. In assays of replated primary or secondary colonies, inclusion of irradiated marrow-cell feeders was found to be an additional requirement. The mixed erythroid-megakaryocyte-granulocyte nature of colonies identified simply as macroscopic and erythroid was confirmed by cytochemical stains for lineage-specific markers. Marked variation in self-renewal was a feature of marrow stem cells both before and after maintenance in flask culture, although the overall self-renewal capacity exhibited by flask-cultured cells was approximately 5-fold higher. Variation in self-renewal was not correlated with primary colony size, which also varied over a wide range (0.2-9 X 10(5) nucleated cells per colony). Variation in stem cell self-renewal has been previously associated with hemopoietic stem cell proliferation in vivo. Its persistence in vitro in assays of dilute single-cell suspensions casts doubt on the significance of microenvironmental influences in directing stem cell differentiation.