High-grade gliomas remain incurable and lethal. Through the availability of the stem-like cells responsible for glioblastoma (GB) formation, expansion, resilience and recurrence, the discovery of glioma cancer stem cells (GCSCs) is revolutionizing this field. GCSCs provide an unprecedented opportunity to reproduce and study GB pathophysiology more accurately. This critically emphasizes our ability to unambiguously identify, isolate and investigate cells that do qualify as GCSCs, to use them as a potential model that is truly predictive of GBs and of their regulation and response to therapeutic agents. We review this concept against the background of key findings on somatic, neural and solid tumour stem cells (SCs), also taking into account the emerging phenomenon of phenotypic SC plasticity. We suggest that basic approaches in these areas can be imported into the GCSC field, so that the same functional method used to identify normal somatic SCs becomes the most appropriate to define GCSCs. This, combined with knowledge of the cellular and molecular basis of normal adult neurogenesis, promises to improve the identification of GCSCs and of selective markers, as well as the development of innovative, more specific and efficacious antiglioma strategies.