Although there is evidence for correlations between disability and magnetic resonance imaging (MRI) total lesion volume in autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), the significance of structural MRI abnormalities for cognitive dysfunction remains controversial. We performed detailed neuropsychological testing, high resolution MRI, and Tc-99m-ethyl cysteinate-dimer SPECT in three CADASIL patients. MR-images were rated independently by two investigators for the presence of white matter lesions, lacunar infarcts, microbleeds, and ventricular enlargement. Cortical atrophy was quantified by the use of automatic morphometric assessment of the cortical thickness. In addition, laboratory and patients' history data were collected in order to assess the individual vascular risk factor profile. The differences in cognitive performance between the three patients are neither explained by structural-, or functional neuroimaging, nor by the patient-specific vascular risk factor profiles. The neuroradiologically least affected patient met criteria for dementia, whereas the most severely affected patient was in the best clinical and cognitive state. Conventional structural and functional neuroimaging is important for the diagnosis of CADASIL, but it is no sufficient surrogate marker for the associated cognitive decline. Detailed neuropsychological assessment seems to be more useful, particularly with respect to the implementation of reliable outcome parameters in possible therapeutic trials.