Glioblastoma is the most frequent primary brain tumor in the adult, accounting for 53.8% of all gliomas (http://www.cbtrus.org), and it is one of the most deadly among all human tumors. Despite aggressive treatment at diagnosis, consisting of resection followed by radiation with concurrent and subsequent adjuvant chemotherapy with temozolomide, the tumor almost invariably recurs or progresses, with a patient median survival of 14.6 mo (1). The hallmark of glioblastoma that distinguishes it from all of the other glial tumors is microvascular proliferation in conjunction with necrosis. Therefore, treatment with antiangiogenic agents holds great promise to block the growth of this most vascularized tumor. The best-known antiangiogenic agents are inhibitors of VEGF-A, an indispensable angiogenic factor during developmental organogenesis and growth of numerous tumors. However, treatment with bevacizumab, a neutralizing antibody to VEGF-A, at relapse only confers transient benefit and a marginal increase in survival, indicating at tumor progression a VEGF-independent angiogenic mechanism of glioblastoma resistance (2). Several mechanisms have been implicated in angiogenesis. One is the sprouting of capillaries from preexisting blood vessels by endothelial proliferation (3). Another is the cooption of preexisting blood vessels by tumor cells, leading to expression of angiopoietin-2 by those vessels’ endothelial cells and tumor cell proliferation, followed later by involution of preexisting vessels in the core of the tumor, massive tumor cell apoptosis, organization of remaining tumor cells into pseudopalisading that resides around areas of necrosis, and tumor rescue at the margins by angiogenesis (4, 5). Expression of HIF-1α and up-regulation of VEGF-A have been identified in hypoxic perinecrotic pseudopalisading tumor cells (4–6). Hypoxia induces elevated levels of VEGF-A (6) and VEGF-A receptors that appear up-regulated in tumor endothelial cells but not in normal brain (7). Another mechanism is the release of angiogenic factors by the tumor that recruit bone marrow-derived endothelial progenitors, hematopoietic stem and progenitor cells that participate in vessel formation (8–10). In PNAS, Soda, Verma, and colleagues reveal a new paradigm for glioblastoma angiogenesis whose main contribution is transdifferentiation of glioblastoma cells into endothelial cells (11) (Fig. 1). Notably, these tumor-derived endothelial cells (TDECs) are refractory to inhibition of both VEGF-A and basic fibroblast growth factor (bFGF, FGF-2) pathways. By mapping GFP+ p53-deficient glioblastoma established in glial-specific Cre mice (GFAP-Cre) (12), Soda et al. find that tumor cells can directly transdifferentiate into CD31+CD34+ endothelial cells that lack VEGF-A receptors (VEGFR), constituting over 20% of total CD31+CD34+ tumor endothelial population. These TDECs are capable of forming patent vessels. Moreover, further analysis by hypoxyprobe unraveled their preferential localization in deep hypoxic areas of the tumors. The hypoxic-associated distribution of TDECs and elevated expression of HIF-1α, a hypoxia-induced transcription factor, indicate the role of hypoxia as the key determinant in forcing putative glioma cells to differentiate into endothelial-like cells.