Therapies directed against VEGF-A and its receptors are effective in treating many mouse tumors but have been less so in treating human cancer patients. To elucidate the reasons that might be responsible for this difference in response, we investigated the nature of the blood vessels that appear in human and mouse cancers and the tumor "surrogate" blood vessels that develop in immunodeficient mice in response to an adenovirus expressing VEGF-A(164). Both tumor and tumor surrogate blood vessels are heterogeneous and form by two distinct processes, angiogenesis and arterio-venogenesis. The first new angiogenic blood vessels to form are mother vessels (MV); MV arise from preexisting venules and capillaries and evolve over time into glomeruloid microvascular proliferations (GMP) and subsequently into capillaries and vascular malformations (VM). Arterio-venogenesis results from the remodeling and enlargement of preexisting arteries and veins, leading to the formation of feeder arteries (FA) and draining veins (DV) that supply and drain angiogenic vessels. Of these different blood vessel types, only the two that form first, MV and GMP, were highly responsive to anti-VEGF therapy, whereas "late"-formed capillaries, VM, FA and DV were relatively unresponsive. This finding may explain, at least in part, the relatively poor response of human cancers to anti-VEGF/VEGFR therapies, because human cancers, present for months or years prior to discovery, are expected to contain a large proportion of late-formed blood vessels. The future of anti-vascular cancer therapy may depend on finding new targets on "late" vessels, apart from those associated with the VEGF/VEGFR axis.