Glioblastomas are a highly aggressive brain tumor, with one of the highest rates of new blood vessel formation. In this study we used a combined experimental and bioinformatics strategy to determine which genes were highly expressed and specific for glioblastoma endothelial cells (GBM-ECs), compared to gene expression in normal tissue and endothelium. Starting from fresh glioblastomas, several rounds of negative and positive selection were used to isolate GBM-ECs and extract total RNA. Using Serial Analysis of Gene Expression (SAGE), 116,259 transcript tags (35,833 unique tags) were sequenced. From this expression analysis, we found 87 tags that were not expressed in normal brain. Further subtraction of normal endothelium, bone marrow, white blood cell and other normal tissue transcripts resulted in just three gene transcripts, ANAPC10, PLXDC1(TEM7), and CYP27B1, that are highly specific to GBM-ECs. Immunohistochemistry with an antibody for PLXDC1 showed protein expression in GBM microvasculature, but not in the normal brain endothelium tested. Our results suggest that this study succeeded in identifying GBM-EC specific genes. The entire gene expression profile for the GBM-ECs and other tissues used in this study are available at SAGE Genie (http://cgap.nci.nih.gov/SAGE). Functionally, the protein products of the three tags most specific to GBM-ECs have been implicated in processes critical to endothelial cell proliferation and differentiation, and are potential targets for anti-angiogenesis based therapy.