In the aldehyde-fixed rat brain NADPH-diaphorase is suggested to be related to brain nitric oxide synthase but also to other isoforms of this enzyme as well as to several non-related types of NADPH-oxidoreductases. In this study NADPH-diaphorase histochemistry using the tetrazolium salt BSPT (2-(2'-benzothiazolyl)-5-styryl-3-(4'-phthalhydrazidyl)-tetrazoliu m chloride) (to yield an electron dense formazan) and immunocytochemistry were applied for the cellular and subcellular localization of brain nitric oxide synthase in the striatum and the pontine laterodorsal tegmental nucleus of the rat. Combining the two techniques, in both brain regions identical distribution patterns of heavily-stained neurons were observed at the light microscopic level. There are inconsistencies in the literature with regard to the subcellular localization of brain nitric oxide synthase and NADPH-diaphorase in neurons. In our results brain nitric oxide synthase immunoreactivity in abundantly stained neurons was mainly cytosolically distributed, sometimes in a patch-like form and distant from membranes, whereas the NADPH-diaphorase reaction product BSPT-formazan was closely attached to discrete portions of intracellular membranes. Other neurons and glial cells including their processes showed also, but to a lesser extent, formazan-labelled membrane portions. In such cell populations brain nitric oxide synthase immunoreactivity was not detectable. Possible reasons for these inconsistencies are discussed in detail. The strength but not the specificity of the NADPH-diaphorase related reaction was shown to be dependent on concentrations of Triton X-100 and tetrazolium salt. We suggest that, for electron microscopical cytochemistry, the BSPT technique combined with other independent techniques, such as immunocytochemistry and in situ hybridization, may be a viable means for the identification and subcellular localization of the different nitric oxide synthase isoforms, and to discriminate them from other types of NADPH-diaphorases.