Relatively new developments in MRI, such as functional MRI (fMRI), magnetic resonance spectroscopy (MRS) and diffusion tensor imaging (DTI) are rapidly developing into imaging modalities that will become clinically available in the near future. They have in common that their signal is somewhat easier to interpret than structural MRI: fMRI mirrors excess cerebral blood flow, in many cases representing brain activity, MRS gives the average volume concentrations of specific chemical compounds, and DTI reflects "directedness" of micro-anatomical structures, of particular use in white matter where fiber bundle disruption can be detected with great sensitivity. While structural changes in MRI have been disappointing in giving a diagnosis of sufficient sensitivity and specificity, these newer methods hold out hope for elucidating pathological changes and differentiating patient groups more rigorously. This paper summarizes promising research results that will yet have to be translated into real life clinical studies in larger groups of patients (e.g. memory clinic patients). Where available, we have tried to summarize results comparing different types of dementia.