It has been established that nerve cell bodies of the nucleus raphe dorsalis (NRD) belong to ascending 5-hydroxytryptamine systems. These neurons could be modulated by GABAergic interneurons or interposed GABA neurons. A high glutamate decarboxylase (GAD) activity in the NRD and a specific high-affinity uptake mechanism for GABA suggest the presence of GABA synthesizing elements in the NRD. Anti-GAD antibodies were used by an immunocytochemical procedure to demonstrate the presence of GABAergic elements. Anti-GAD antibodies were previously tested in the cerebellum and substantia nigra. Large amounts of GAD-positive reaction product were observed in the cytoplasm of some neurons (fusiform, ovoid or multipolar) or appeared as punctate deposits apposed to dendrites, soma and dispersed in the neuropil of the NRD. At the electron microscopic level, GAD-positive reaction product was observed within the cytoplasm of numerous somata in sections from colchicine-treated rats. GAD-positive staining was observed in numerous fibers or axonal terminals and two types of morphologically different fibers could be distinguished. The first displays small clear vesicles and few large granular vesicles (LGV) (80-100 nm), the second displays only clear round vesicles (40-60 nm). After 5,7-dihydroxytryptamine treatment (a neurotoxic for 5-HT terminals), the immunocytochemical labeling is much decreased. Some reactive neurons are still dispersed in the nucleus but the fibers containing LGV are no longer observed. These results strongly suggest that some neuronal elements in the NRD are morphologically, pharmacologically and anatomically similar to 5-HT neurons described at this level. Such cell elements could possess a double GABA and 5-HT potentiality. If this is not the case, a population of GABA neurons could be sensitive to 5,7-DHT and so have the capacity to take up 5-HT. The other reactive elements, insensitive to 5,7-DHT, could represent the GABAergic interneurons postulated at this level. Numerous GAD positive fibers or axon terminals were observed in synaptic contact with dendrites, axons or soma of other neurons. The chemical nature of the neuronal postsynaptic elements remains unknown. These findings strongly support the hypothesis for GABA-mediated inhibition in the NRD.