Amyloid-β-protein (Aβ) is generally assessed by neuropathologists in diagnostics. This BrainNet Europe ( http://www.brainnet-europe.org/ ) (15 centres and 26 participants) study was carried out to investigate the reliability of such an assessment. In the first part of this trial, tissue microarray sections were stained with the antibody of each centre’s choice. Reflecting the reality, seven antibodies and a plethora of pretreatment strategies were used. Ninety-two percent of the stainings were of good/acceptable quality and the estimation of presence of Aβ aggregates yielded good results. However, a poor agreement was reached particularly regarding quantitative (density) and qualitative (diffuse/cored plaques) results. During a joint meeting, the clone 4G8 was determined to label best the fleecy/diffuse plaques, and thus, this clone and the formic acid pretreatment technique were selected for the second part of this study. Subsequently, all stained sections were of good/acceptable quality and again a high level of concordance of the dichotomized (presence/absence) assessment of plaques and CAA was achieved. However, even when only one antibody was used, the type of Aβ-aggregates (diffuse/cored), type of vessel and Vonsattel grade, were not reliably assigned. Furthermore, the quantification of lesions was far from reliable. In line with the first trial, the agreement while assessing density (some, moderate and many) was unimpressive. In conclusion, we can confirm the utility of immunohistochemical detection of Aβ-protein in diagnostics and research. It is noteworthy that to reach reproducible results a dichotomized assessment of Aβ-immunoreactivity rather than quantification and assignment of various types of lesions should be applied, particularly when comparing results obtained by different neuropathologists.