We have performed a quantitative analysis of the amyloid load (plaques), neuritic plaques and neurofibrillary tangles (NFT) in the frontal, temporal and parietal association cortices of autopsied brains from 49 prospectively evaluated patients with Alzheimer's disease (AD) diagnosed according to three sets of published pathological criteria. These patients had been assessed clinically with psychological testing of cognitive abilities within 6 months of death. Correlations were sought between severity of pathological change and cognitive status before death, duration of disease and age at death. Using Khachaturian and CERAD criteria highly positive correlations were obtained between the extent of cognitive deficit and the density of NFT in frontal and parietal lobes. The percentage area of cortex occupied by amyloid in the parietal lobe was correlated to the cognitive deficit only in the CERAD-diagnosed cases. The density of all amyloid plaques (AP) showed no correlation with the extent of cognitive deficit, but the densities of neuritic plaques did correlate with cognitive deficit. Both amyloid load and tangle densities were positively correlated with disease duration. All these correlations were reduced or absent in a subgroup of cases fulfilling the Tierney et al. A3 diagnostic criteria for AD. We found no pathological measure that correlated with the age of patients at death. Amyloid loads and NFT densities showed highly significant but selective positive correlations, the most striking being between temporal lobe NFT density and frontal and parietal lobe amyloid load and between temporal lobe NFT density and frontal and parietal lobe NFT densities. Correlations involving AP density as a measure of amyloid load were almost always less significant than those involving the percentage area of cortex occupied by amyloid, suggesting that the latter measures amyloid load more accurately. However, the highest correlations of NFT densities were with neuritic plaque densities. Overall this study highlights the relevance of neuritic changes (revealed by NFT and neuritic plaques) and the irrelevance of amyloid plaques to the dementia of AD.