Résumé : The study reported was intended to compare the impressions and analyses of investigators from 11 different laboratories on 2 slides, each from 6 cases with varying quantities of neuropathological change of the type found in Alzheimer's disease and normal ageing. The material came from 6 selected female patients over 75 years of age all of whom had been examined in detail and assessed by the Blessed Test Score. Two were severely demented, 2 mildly demented and 2 were considered to be normal. Unstained paraffin-embedded slides were sent to the investigators, the choice of the staining techniques being left to each laboratory. A quantitative evaluation of the changes was requested in 2 specified areas of the hippocampus and in the first temporal gyrus. Subjective scores of severity and a final guess about the pre mortem intellectual status (demented or not) were asked. The 11 replies were analyzed. A total of 14 different staining techniques were used. Absolute values of density differed much from one investigator to another, for senile plaques as well as for neurofibrillary tangles. Statistical analysis showed that concordance might be improved by the use of corrective factors which would standardize the scales of measurement. The ranking of the slides in increasing order of severity was in good agreement for 9 out of 11 observers concerning the neurofibrillary tangles and 3 out of 9 observers concerning the senile plaques. The correlation between the intellectual status and the density of lesions was higher for neurofibrillary tangles than for senile plaques. The subjective scores were in better agreement for the severely affected cases than for the mildly affected ones. The lowest correlation with intellectual deficit was obtained with the quantitative scores which took into account only the senile plaques or only the hippocampal lesions. The highest correlation coefficients were obtained with the subjective scores. The observers guessed correctly the intellectual status of the 2 most affected cases and often disagreed for the intermediate and normal cases. Neuropathology is mandatory for the diagnosis of definite Alzheimer's disease. Quantitative assessment is useful in cases with few lesions and light dementia but the neuropathological diagnostic procedure has to be more strictly standardized before quantitative histopathological criteria can be reliably transferred from one laboratory to another, especially when mildly affected cases are involved. Concordance seems presently easier to obtain by ranking the lesions and the cases in increasing order of severity than by using quantitative values of density.(ABSTRACT TRUNCATED AT 400 WORDS)