Various cytoskeletal proteins have been implicated in the cellular pathology of Alzheimer's disease. A monoclonal antibody (SMI32) that recognizes nonphosphorylated epitopes on the medium (168 kDa) and heavy (200 kDa) subunits of neurofilament proteins has been used to label and analyze a specific subpopulation of pyramidal neurons in the prefrontal and inferior temporal cortices of normal and Alzheimer's disease brains. In Alzheimer's disease, the distribution of neuropathological markers predominates in layers III and V in these association areas. In these neocortical regions, SMI32 primarily labels the perikarya and dendrites of large pyramidal neurons, predominantly located within layers III and V. In Alzheimer's disease, a dramatic loss of SMI32-immunoreactive (ir) cells was observed, affecting particularly the largest cells (i.e., cells with a cross-sectional perikaryal area larger than 350 microns 2). The staining intensity of the largest SMI32-ir neurons was significantly reduced in Alzheimer's disease cases, suggesting that an inappropriate phosphorylation of these cytoskeletal proteins may take place in the course of the pathological process. In addition, the SMI32-ir neuron loss and total neuron loss were highly correlated with neurofibrillary tangle counts, whereas such a correlation was not observed with neuritic plaque counts. These quantitative data suggest that SMI32-ir neurons represent a small subset of pyramidal cells that share certain anatomical and molecular characteristics and are highly vulnerable in Alzheimer's disease. Other studies have suggested that SMI32-ir neurons are likely to furnish long corticocortical projections. Thus, their loss would substantially diminish the effectiveness of the distributed processing capacity of the neocortex, resulting in a neocortical isolation syndrome as reflected by the clinical symptomatology observed in these patients. Such correlations between the expression of a selective cellular pathology and specific elements of cortical circuitry will increase our understanding of the molecular and cellular characteristics underlying a given neuronal subclass vulnerability in Alzheimer's disease or other neurodegenerative disorders.