The purpose of this work was to establish a neuropathological confirmation of the clinical diagnosis of senile dementia of the Alzheimer type (SDAT) in a group of patients prospectively studied in a geriatric hospital since 1984 (Charles Richet Study). The sample consisted of 45 cases, 35 of which had received a clinical diagnosis of SDAT and 10 others a diagnosis of either vascular or mixed dementia. The mean age at death was 85 +/- 6.9 years (range 64-97). The neuropathological diagnosis was established independently of the clinical findings. Senile lesions typical of SDAT were found in 32/35 cases, giving a 91.4% rate of clinico-pathologic agreement. However, vascular lesions were present in 10 cases (28.5%) and the final pathologic diagnosis was mixed dementia, lowering the score of agreement for SDAT to 63%. Considering that the pathological criteria for the diagnosis of SDAT are not uniform, we independently applied 3 inclusion (senile lesions) and 3 exclusion (vascular lesions) criteria for the diagnosis of SDAT to each of the 45 cases. This permitted a comparison of nine combinations of neuropathological criteria with the clinical diagnosis. The sensitivity values ranged from 37 to 80% and the specificity values ranged from 55 to 100%. The highest agreement rate with the clinical diagnosis was achieved when were associated the criterion which specified that plaques and tangles must be present in the hippocampus regardless of neocortical findings and the criterion which excluded cases with vascular lesions of any site if their volume was 50 ml or more.(ABSTRACT TRUNCATED AT 250 WORDS)