Neuronal nitric oxide synthase (nNOS) mRNA levels and NADPH diaphorase (NADPH-d) staining were compared in the frontal cortex, visual cortex and hippocampus (dentate gyrus and CA subfields of Ammon's horn) of five Alzheimer's disease (AD) and six control brains. The cellular abundance of nNOS mRNA was quantified by in-situ hybridisation using 35S-labelled antisense oligonucleotides complementary to the human nNOS sequence. Although the mean level of nNOS expression was decreased in all three regions in AD cases as compared to controls, it did not reach significance. Neurones positively labelled for nNOS mRNA and neurones positive for NADPH-d histochemistry displayed similar distribution in control and AD cases. In AD brains the density of neurones having detectable levels of nNOS mRNA was significantly decreased in the white matter underlying the frontal cortex (P < 0.05) but not in the frontal cortex gray matter; no change was observed in the gray or white matter of the visual cortex in AD. The number of cells expressing detectable levels of nNOS mRNA in the hippocampus was also significantly decreased (P < 0.05) in AD. The density of NADPH-d-positive cells was not significantly decreased in the gray or white matter of the frontal or visual cortices in AD compared to controls; however, the number of NADPH-d-positive cells was significantly decreased in the hippocampus (P < 0.01). These data indicate that although the cellular abundance of nNOS mRNA is not significantly decreased in these three regions in AD, there is a significant decrease in the number of cells expressing detectable levels of nNOS mRNA in the white matter underlying the frontal cortex and in the dentate gyrus and CA subfields of the hippocampus in AD. Furthermore, there was also a significant decrease in the number of NADPH-d-positive cells in the dentate gyrus and CA subfields of the hippocampus in AD as compared to controls. These results suggest specific populations of nNOS/NADPH-d cells in the white matter underlying the frontal cortex and in the hippocampus are vulnerable in AD. The implications of these findings are discussed.