Recently, as in many solid tumors, cancer stem cells (CSC) or tumor progenitor cells have been defined as the probable source of human high-grade gliomas, including the most malignant type, glioblastoma. CSCs can be isolated from most of the primary glioblastomas using adequate culture conditions and show features of true stem cells like infinite self-renewal, clonal properties and differentiation in several lines of differentiated cells. There is some evidence that CSCs develop from neural stem cells (NSC), though definite proof for this assumption is lacking at this point. However, there is much evidence that CSCs can only develop in a proper microenvironment, the tumor niche. This tumor niche shares many properties with a similar niche, the stem cell niche, where NSCs are situated. Any kind of a physiological or pathological niche consists of several cellular components (neurons, astroglia, oligodendrocytes, endothelial cells, microglia, immune cells and several progenitor cells) that secrete soluble factors of different origin, and the extracellular matrix. In addition, the regional metabolism plays an important role that has been underestimated until recently. The so called Warburg effect describes a phenomenon of aerobic glycolysis found in many solid tumors, including glioblastomas, which has not been completely understood to date, but which might explain the transition from a normal NSC niche to a pathological CSC niche. It seems plausible that CSCs develop in such a niche and re-generate the tumor bulk that mainly consists of differentiated tumor cells. This would further explain the failure of many of the classical therapies that mostly target differentiated tumor cells. The talk will summarize the available evidence for the outlined hypothesis and provide strategies for the future investigation of the phenomenon.