Quantitative clinicopathological correlation studies are one way to address the question of the relevance of morphological abnormalities in Alzheimer's dementia (AD). This paper summarizes results of the Vienna Longitudinal Study on Dementia obtained during the past few years and presents a critical discussion on the relevance of clinicopathological correlation studies for the pathogenesis of AD. Plotting of psychometric test scores against the numbers of plaques, tangles and neuropil threads in various cortical areas shows that significant correlations are due primarily to very high lesion counts in severely demented patients. These data indicate that neocortical neurofibrillary pathology can be considered an end-stage marker in the pathology of AD. On the other hand, the topographical staging of neuritic Alzheimer changes proposed by Braak and Braak (1991) appears to be a better reflection of the progression of the degenerative process than numerical lesion counts; there is a linear correlation between the Braak stages and Mini-Mental State scores in 122 aged individuals. Significant correlations are further obtained between the severity of dementia and the levels of a number of synaptic proteins including synaptophysin and the chromogranins. Taken together, our data suggest that none of the classical AD lesions, plaques and tangles, play a central role in the pathogenesis of dementia, a fact that is supported by a molecular biological study showing that there is no close relationship between these lesions and the neurons undergoing degeneration in AD. Whereas neuritic pathology is a useful histopathological marker for the diagnosis and staging of AD, the major correlate of cognitive deficits is the loss of corticocortical and subcorticocortical connections reflected by a depletion of synapses. This pathology may be induced by a mismetabolism of the beta-amyloid precursor proteins or their interaction with cytoskeletal proteins related to neuronal degeneration.