The neuropathological diagnosis of Alzheimer's disease requires an assessment of the quantity of pathology present. Advances in molecular biology have highlighted the role of beta-amyloid precursor protein (beta APP) in the pathogenesis of the disease. This protein is found in neurons and other cells and many neuropathological studies would benefit from a method which generates reliable data on the numbers of cells containing significant amounts of the protein. Classically, generation of such data would have involved laborious manual counting. This particular approach carries low levels of inter- and intra-rater reliability and is much dependent on the skill and experience of the operator. We have used immunocytochemistry to specifically define a single cell population, pre-alpha cells, containing beta APP, and have developed a computerized cell counting programme that can reliably quantify these cells in human post-mortem brain samples. We have obtained a high level of accuracy (> 95%) and efficiency in identifying and quantifying target cells and have demonstrated that our protocol can be used effectively by both novice and expert. This method could be easily configured to provide quantitative data for a wide range of immunocytochemically defined cell populations.