Malignant primary brain tumors are characterized by a short median survival and an almost 100% tumor-related mortality. Despite the addition of new chemotherapy regimes, the overall survival has improved marginally, and radiotherapy is only transiently effective, illustrating the profound impact of treatment resistance on prognosis. Recent studies suggest that a small subpopulation of cancer stem cells (CSCs) has the capacity to repopulate tumors and drive malignant progression and mediate radio- and chemoresistance. This implies that future therapies should turn from the elimination of the rapidly dividing, but differentiated tumor cells, to specifically targeting the minority of tumor cells that repopulate the tumor. Although there exists some support for the CSC hypothesis, there remain many uncertainties regarding theoretical, technical, and interpretational aspects of the data supporting it. If correct, the CSC hypothesis could have profound implications for the way tumors are classified and treated. In this review of the literature, we provide original data and hypotheses supporting alternative explanations and outline some of the therapeutic implications that can be derived.