We studied the growth-associated protein, GAP-43 (also called neuromodulin and B-50) in post-mortem brain tissue using immunohistochemistry and quantitative Western blotting, from patients with Alzheimer's disease (AD) and age-matched control subjects. By immunohistochemistry, we found a clear reduction of GAP-43 in the frontal cortex, while in the hippocampus, there was a marked reduction in some areas (dentate molecular layer, stratum moleculare and radiale of CA1 and CA4), while not in other areas (stratum lacunosum, pyramidale and oriens of CA1). Moreover, in the hippocampus, neuritic staining was prominent, and was often associated with senile plaques. Quantitative analysis showed that GAP-43 was significantly reduced in AD, both in the frontal cortex (70% of the control value, p < 0.01) and in the hippocampus (81% of the control value, p < 0.05). In the frontal cortex, there was a significant negative correlation between GAP-43 and duration of dementia (r = -0.58; p < 0.02) and a positive correlation between GAP-43 and the synaptic vesicle-specific protein rab3a (r = 0.62; p < 0.05), while no such correlation were found in the hippocampus. In contrast, a significant positive correlation was found between GAP-43 and the number of senile plaques in the hippocampus (r = 0.64; p < 0.05), but not in the frontal cortex. GAP-43 is known to be involved in maintenance of synapses and in neuritic regeneration. Our findings may suggest that in the frontal cortex, GAP-43 levels decline as a consequence of the synaptic degeneration, while in the hippocampus, sprouting processes, involving GAP-43, are active.