For more than a decade the 'neurosphere assay' has been used to define and measure neural stem cell (NSC) behavior, with similar assays now used in other organ systems and in cancer. We asked whether neurospheres are clonal structures whose diameter, number and composition accurately reflect the proliferation, self-renewal and multipotency of a single founding NSC. Using time-lapse video microscopy, coculture experiments with genetically labeled cells, and analysis of the volume of spheres, we observed that neurospheres are highly motile structures prone to fuse even under ostensibly 'clonal' culture conditions. Chimeric neurospheres were prevalent independent of ages, species and neural structures. Thus, the intrinsic dynamic of neurospheres, as conventionally assayed, introduces confounders. More accurate conditions (for example, plating a single cell per miniwell) will be crucial for assessing clonality, number and fate of stem cells. These cautions probably have implications for the use of 'cytospheres' as an assay in other organ systems and with other cell types, both normal and neoplastic.