The amyloid beta-protein (Abeta) is the main component of Alzheimer's disease-related senile plaques. Although Abeta is associated with the development of Alzheimer's disease, it has not been shown which forms of Abeta induce neurodegeneration in vivo and which types of neurons are vulnerable. To address these questions, we implanted DiI crystals into the left frontocentral cortex of APP23 transgenic mice overexpressing mutant human APP (amyloid precursor protein gene) and of littermate controls. Traced commissural neurons in layer III of the right frontocentral cortex were quantified in 3-, 5-, 11- and 15-month-old mice. Three different types of commissural neurons were traced. At 3 months of age no differences in the number of labelled commissural neurons were seen in APP23 mice compared with wild-type mice. A selective reduction of the heavily ramified type of neurons was observed in APP23 mice compared with wild-type animals at 5, 11 and 15 months of age, starting when the first Abeta-deposits occurred in the frontocentral cortex at 5 months. The other two types of commissural neurons did not show alterations at 5 and 11 months. At 15 months, the number of traced sparsely ramified pyramidal neurons was reduced in addition to that of the heavily ramified neurons in APP23 mice compared with wild-type mice. At this time Abeta-deposits were seen in the neo- and allocortex as well as in the basal ganglia and the thalamus. In summary, our results show that Abeta induces progressive degeneration of distinct types of commissural neurons. Degeneration of the most vulnerable neurons starts in parallel with the occurrence of the first fibrillar Abeta-deposits in the neocortex, that is, with the detection of aggregated Abeta. The involvement of additional neuronal subpopulations is associated with the expansion of Abeta-deposition into further brain regions. The vulnerability of different types of neurons to Abeta, thereby, is presumably related to the complexity of their dendritic morphology.