From a clinical as well as a neuropathological point of view Alzheimers disease (AD) has been the focus of intense research for more than three decades. Most studies to identify morphometric correlates with the declining cognitive function in normal aging and AD have employed semi-quantitative methods to assess neuropathological markers such as neurofibrillary tangles, senile plaques, neuronal, or glial cell densities, and neuron sizes. To this end, many cell counting methods have employed two-dimensional designs in single sections, yielding estimates of cell numbers either as neuron densities (number of cell profiles per area) or estimates of the size distribution of neuron profiles in columns vertical to the cortical surface. This approach gives rise to difficulties in interpretation because of the three-dimensional size, shape, and orientation of the counted cells, and the effect of shrinkage artifacts. Modern stereological techniques offer a more rigorous approach for quantifying neuropathological changes associated with aging and degenerative disease. However the stereological studies also suffer from the limitations of high biological variability in AD-type neuropathology, and the relative scarcity of autopsied brains from well-studied non-demented comparison subjects. As a result, the clinicopathological associations between neuropathology and indices of cognitive performance in aging and AD are not yet firmly established. The requirement for the proper description of morphologic neuropathology of AD is clear: any macroscopic or microscopic abnormalities, are subtle and must consequently be demonstrated reproducibly in well-controlled studies. In this review we try to evaluate which, if any, of the contemporary claims for morphometric brain abnormalities in AD can be said to be well established, with special emphasis placed on human stereological post-mortal studies.