Glioblastoma is one of the most aggressive brain tumours, with a life expectancy of 12-15 months. Structural MRI lacks sensitivity for detecting early therapeutic response. DWI is sensitive to changes in cellular density and provides a marker of cell kill. There is little published data on which of the diffusion components detectable in diseased brain tissue in vivo is most sensitive to this. Voxel by voxel pre- and post-treatment analysis in ‘functional diffusion map’ studies are confounded by change in tumour size. We assess the utility of a spatially-independent histogram approach to detect early treatment response using advanced DWI models.