GABAergic neurons in laminae I-III of the spinal dorsal horn may contain one or more of the following compounds: glycine, acetylcholine, neuropeptide Y, enkephalin, nitric oxide synthase or parvalbumin. Although the pattern of co-localization of some of these compounds is understood, it is not known which types of GABAergic neurons contain parvalbumin, or whether nitric oxide synthase coexists with peptides, acetylcholine or parvalbumin in any of these neurons, and in this study we have used immunocytochemistry and enzyme histochemistry to resolve these issues. Parvalbumin-immunoreactivity was restricted to those GABA-immunoreactive neurons that also showed glycine-immunoreactivity and was not co-localized with neuropeptide Y-immunoreactivity or NADPH diaphorase activity. By combining NADPH diaphorase histochemistry with immunocytochemistry with an antiserum to nitric oxide synthase, we were able to show that NADPH diaphorase activity was a reliable marker for nitric oxide synthase in the spinal cord. Neurons that possess GABA- but not glycine-immunoreactivity may contain neuropeptide Y, enkephalin, acetylcholine or NADPH diaphorase, and all of the cholinergic neurons appear to contain NADPH diaphorase. By combining immunofluorescent detection of neuropeptide Y or enkephalin with NADPH diaphorase histochemistry, we showed that peptide-immunoreactivity did not coexist with NADPH diaphorase. This suggests that neither of these peptides coexists with nitric oxide synthase or with acetylcholine in neurons in the superficial dorsal horn. Several phenotypically distinct groups of GABA-immunoreactive neuron can therefore be identified in laminae I-III of the dorsal horn, and these may represent different functional types of inhibitory neuron.