Glioblastoma multiforme (GBM) is the most aggressive form of brain tumors with no curative treatments available to date. Murine models of this pathology rely on the injection of a suspension of glioma cells into the brain parenchyma following incision of the dura-mater. Whereas the cells have to be injected superficially to be accessible to intravital two-photon microscopy, superficial injections fail to recapitulate the physiopathological conditions. Indeed, escaping through the injection tract most tumor cells reach the extra-dural space where they expand abnormally fast in absence of mechanical constraints from the parenchyma. Our improvements consist not only in focally implanting a glioma spheroid rather than injecting a suspension of glioma cells in the superficial layers of the cerebral cortex but also in clogging the injection site by a cross-linked dextran gel hemi-bead that is glued to the surrounding parenchyma and sealed to dura-mater with cyanoacrylate. Altogether these measures enforce the physiological expansion and infiltration of the tumor cells inside the brain parenchyma. Craniotomy was finally closed with a glass window cemented to the skull to allow chronic imaging over weeks in absence of scar tissue development. Taking advantage of fluorescent transgenic animals grafted with fluorescent tumor cells we have shown that the dynamics of interactions occurring between glioma cells, neurons (e.g. Thy1-CFP mice) and vasculature (highlighted by an intravenous injection of a fluorescent dye) can be visualized by intravital two-photon microscopy during the progression of the disease. The possibility to image a tumor at microscopic resolution in a minimally compromised cerebral environment represents an improvement of current GBM animal models which should benefit the field of neuro-oncology and drug testing.