Vascular lesions are particularly common in aged brain. However, it is still unclear whether all such lesions affect cognition. Methodological problems: masking effect of concomitant pathologies, heterogeneity of cerebral vascular lesions, highly variable location and lack of a clear cut pathologic correlate of certain neuroradiological findings may explain discrepancies among published studies. In order to better explore relationships between specific characteristics of vascular lesions (type, size and location) and cognitive status we performed several clinicopathological studies in relatively "pure" series of autopsied elderly individuals with varying levels of cognitive impairment. Cognitive status was classified according to the Clinical Dementia Rating (CDR) scale and neuropathological evaluation included A_-protein deposition and neurofibrillary tangle staging and bilateral semiquantitative assessment of ischemic lesions. In a first series of 45 cases without significant neurofibrillary tangle pathology or macrovascular lesions, we report that cortical microinfarcts explained 36.1% of the variability in CDR, in contrast to focal cortical and white matter gliosis, which were not significantly associated with CDR. In a second series of 72 cases without significant neurofibrillary tangle pathology and without macroscopic ischemic lesions other than lacunes, we found that basal ganglia and thalamic lacunes, periventricular and diffuse white matter demylelination explained 6%, 5%, 6% and 6% of the CDR variability, respectively. Lacunes in the frontal, temporal and parietal deep white matter did not correlate with cognitive decline in this series. Microscopic ischemic pathology, basal ganglia and thalamic lacunes and both diffuse and periventricual demyelination should be taken into account for the elaboration of future neuropathological criteria for vascular and mixed dementia.