Tumor vasculature is in general highly heterogeneous. This characteristic is most prominent in high-grade gliomas, which present with areas of angiogenic growth, next to large areas of diffuse infiltrative growth in which tumor cells thrive on pre-existent brain vasculature. This limits the effectiveness of anti-angiogenic compounds as these will not affect more matured and co-opted vessels. Therefore, additional destruction of existing tumor vasculature may be a promising alternative avenue to effectively deprive tumors from blood. This approach requires the identification of novel tumor vascular targeting agents, which have broad tumor vessel specificities, ie are not restricted to newly formed vessels. Here, we describe the generation of a phage library displaying nanobodies that were cloned from lymphocytes of a Llama which had been immunized with clinical glioma tissue. In vivo biopanning with this library in the orthotopic glioma xenograft models E98 and E434 resulted in the selection of various nanobodies which specifically recognized glioma vessels in corresponding glioma xenografts. Importantly, also nanobodies were isolated which discriminated incorporated pre-existent vessels in highly infiltrative cerebral E434 xenografts from normal brain vessels. Our results suggest that the generation of nanobody-displaying immune phage libraries and subsequent in vivo biopanning in appropriate animal models is a promising approach for the identification of novel vascular targeting agents.