Certain cells (CFU-S) in the haematopoietic tissues of mice are able to form macroscopic nodules in the spleen 7–14 days after injection in heavily irradiated recipient mice¹. At 7–9 days, most of the spleen colonies contain recognizable cells of one predominant haematopoietic lineage², and do not contain cells capable of spleen colony formation on injection in secondary hosts³. However, by 14 days most of the spleen colonies do contain cells of more than one line of haematopoietic differentiation², as well as precursor cells capable of again generating similar multilineal spleen colonies on retransplantation³. These observations have led to the widely accepted conclusion that most spleen colonies are derived from pluripotential ‘stem’ cells³, and to the suggestion that the apparent transformation in character of spleen colonies from unilineal at early times to mixed later on reflects the local operation of ‘instructive’ differentiative signals in the splenic environment². The reasoning behind these suggestions implicitly assumed that late colonies simply represented a further stage in the development of the same colonies observable at earlier times. Here, we present direct evidence against this assumption. Our data indicate that most spleen colonies identified as surface nodules at days 7–8 are neither multipotential nor self-maintaining, but rather are destined to disappear from the spleen within 72 h. The observations set limits on the validity of the spleen colony method as an assay for self-maintaining pluripotential progenitor cells, and give cause for reassessment of the data on which widely accepted views concerning the regulation of such cells are based.