Human glioblastoma multiforme (GBM) is one of the most devastating cancers.The hallmark of GBM is the invasiveness of the tumor cells infiltrating into normal brain parenchyma, making it virtually impossible to remove the tumor completely by surgery and inevitably leading to recurrent disease.Progress in understanding GBM pathobiology and in developing novel antitumor therapies could be greatly accelerated with animal model systems that display characteristics of human GBM and that enable tumor monitoring through noninvasive imaging in real time.Subjecting human cancer cells to an experimental lung metastasis assay (ELM) often yield highly metastatic cells with higher proliferative and invasive potential.However, the ELM assay has not been tested previously with GBM, most likely because extracranial metastases of human GBM are clinically rare.In this study, we used ELM to enrich metastatic cell populations and found that three of four commonly used GBM lines (U251, U87, and DBTRG-05MG) were highly metastatic after two rounds (M2) of ELM selection.These GBM-M2 lines grew more aggressive orthotopically and all showed significant multifold increases in IL6, IL8, MCP-1, and GM-CSF, which are cytokiues and factors associated with poor GBM prognosis.DBM2 cells, derived from the DBTRG-05MG cell line, not only form highly invasive orthotopic tumors (with areas of central necrosis, vascular hyperplasia, and intracranial dissemination), but also erode the skull, permitting the use of high-resolution micro-ultrasound in real time to non-invasively observe tumor growth and vascularization.In conclusion, commonly used GBM cells have intrinsic metastatic potential which can be selected for in ELM assays.When implanted in the brain, the metastatic potential of GBM cells can be realized as a highly invasive phenotype.The DBM2 mouse model has characteristics that mimic the aggressively invasive behavior of clinical GBM, providing a valuable tool for investigating the factors that modulate glioblastoma growth, assessing invasion and vascularity, and for evaluating novel therapeutic agents in real time.