Neural stem cells (NSCs) are the most primordial and least committed cells of the nervous system, the cells that exist before regional specification develops. Because immunocytochemically-detectable markers that are sufficiently specific and sensitive to define an NSC have not yet been fully defined, we have taken the strong view that, to be termed a "stem cell" in the nervous system--in contrast to a "progenitor" or "precursor" (whose lineage commitment is further restricted)--a single neuroectodermally-derived cell must fulfill an operational definition that is essentially similar to that used in hematopoiesis. In other words, it must possess the following functional properties: (1) "Multipotency", i.e., the ability to yield mature cells in all three fundamental neural lineages throughout the nervous system--neurons (of all subtypes), astrocytes (of all types), oligodendrocytes--in multiple regional and developmental contexts and in a region and developmental stage-appropriate manner. (2) The ability to populate a developing region and/or repopulate an ablated or degenerated region of the nervous system with appropriate cell types. (3) The ability to be serially transplanted. (4) "Self-renewal", i.e., the ability to produce daughter cells (including new NSCs) with identical properties and potential. Having identified a murine neural cell clone that fulfills this strict operational definition--in contrast to other studies that used less rigorous or non-operational criteria for defining an NSC (e.g., the "neurosphere" assay)--we then examined, by comparing gene expression profiles, the relationship such a cell might have to (a) a multipotent somatic stem cell from another organ system (the hematopoietic stem cell [HSC]); (b) a pluripotent stem cell derived from the inner cell mass and hence without organ assignment (an embryonic stem cell); (c) neural cells isolated and maintained primarily as neurospheres but without having been subjected to the above mentioned operational screen ("CNS-derived neurospheres"). ESCs, HSCs, and operationally-defined NSCs--all of which have been identified not only by markers but by functional assays in their respective systems and whose state of differentiation could be synchronized--shared a large number of genes. Although, as expected, the most stem-like genes were expressed by ESCs, NSCs and HSCs shared a number of genes. CNS-derived neurospheres, on the other hand, expressed fewer "stem-like" genes held in common by the other operationally-defined stem cell populations. Rather they displayed a profile more consistent with differentiated neural cells. (Genes of neural identity were shared with the NSC clone.) Interestingly, when the operationally-defined NSC clone was cultured as a neurosphere (rather than in monolayer), its expression pattern shifted from a "stem-like" pattern towards a more "differentiated" one, suggesting that the neurosphere, without functional validation, may be a poor model for predicting stem cell attributes because it consists of heterogeneous populations of cells, only a small proportion of which are truly "stem-like". Furthermore, when operational definitions are employed, a common set of stem-like genes does emerge across both embryonic and somatic stem cells of various organ systems, including the nervous system.