The aim of this thesis was to investigate a nanotechnological approach for the treatment of glioblastoma multiforme (GBM), the most insidious and destructive tumor in the human central nervous system. Firstly, the attention was focused on the development of a model of blood-brain barrier (BBB) as close as possible to the physiological one, exploiting a cellular co-culture of endothelial cells and astrocytes. The ability of lipid magnetic nanovectors (LMNVs) to cross this barrier was assessed, focusing on the mechanism of internalization and on the pathway of transcytosis followed by these particles. Secondly, the number of nanoparticles able to cross the BBB was tested on a 3D model of tumor cell spheroids, in order to evaluate their antiproliferative power.