Purpose: Major shortcomings of traditional mouse models based on xenografted human glioblastoma cell lines are that tumor cells do not invade and that genetic alterations, such as amplification of the epidermal growth factor receptor ( EGFR ) gene, are not maintained. Such models are thus of limited value for preclinical studies. We established a highly invasive model to evaluate the effect of antibodies against EGFR (cetuximab) and vascular endothelial growth factor receptor-2 (antibody DC101). Experimental Design: After short-term culture, glioblastoma spheroids were implanted into the brains of nude mice. Animals were treated either i.c. with cetuximab or i.p. with DC101. Tumor burden was determined histologically using image analysis of 36 different landmark points on serial brain sections. Results: Invasive xenografts were obtained from nine different glioblastomas. Three of seven cases treated with cetuximab responded with significant tumor growth inhibition, whereas four did not. All responsive tumors were derived from glioblastomas exhibiting EGFR amplification and expression of the truncated EGFRvIII variant, which were maintained in the xenografts. All nonresponsive tumors lacked EGFR amplification and EGFRvIII expression. The proportion of apoptotic cells was increased, whereas proliferation and invasion were decreased in responsive tumors. None of four xenograft cases treated with DC101 responded to treatment, and the diffusely invading tumors grew independent of angiogenesis. Conclusions: Inhibition of glioblastoma growth and invasion can be achieved using i.c. delivery of an anti-EGFR antibody, but tumor response depends on the presence of amplified and/or mutated EGFR. Antiangiogenic treatment with DC101 is not effective against diffusely invading tumors.