Postembedding immunogold labeling methods applied to ultrathin and semithin sections of cat dorsomedial medulla showed that neuronal perikarya, dendrites, myelinated and nonmyelinated axons, and axon terminals in the nucleus tractus solitarii contain glycine immunoreactivity. Light microscopic observations on semithin sections revealed that these immunoreactive structures were unevenly distributed throughout the entire nucleus. At the electron microscopic level, synaptic terminals with high levels of glycine-immunoreactivity, assumed to represent those releasing glycine as a neurotransmitter, were discriminated from terminals containing low, probably metabolic levels of glycine-immunoreactivity, by a quantitative analysis method. This compared the immunolabeling of randomly sampled terminals with a reference level of labeling derived from sampling the perikarya of dorsal vagal neurones. The vast majority of these "glycinergic" terminals contained pleomorphic vesicles, formed symmetrical synaptic active zones, and targeted dendrites. They appeared to be more numerous in areas of the nucleus tractus solitarii adjoining the tractus solitarius, but rather scarce caudally, medially, ventrally, and in the dorsal motor vagal nucleus. In a random analysis of the entire nucleus tractus solitarii, 26.2% of sampled terminals were found to qualify as glycine-immunoreactive. In contrast, boutons immunoreactive for gamma-aminobutyric acid (GABA) were more evenly distributed throughout the dorsal vagal complex and accounted for 33.7% of the synaptic terminals sampled. A comparison of serial ultrathin sections suggested three subpopulations of synaptic terminals: one containing high levels of both GABA- and glycine-immunoreactivities (21% of all terminals sampled), one containing only GABA-immunoreactivity (12.7%), and relatively few terminals (5.2%) that were immunoreactive for glycine alone. These results were confirmed by dual labeling of sections using gold particles of different sizes. This study reports the first analysis of the ultrastructure of glycinergic nerve terminals in the cat dorsal vagal complex, and the pattern of coexistence of glycine and GABA observed provides an anatomical explanation for our previously reported inhibitory effects of glycine and GABA on neurones with cardiovascular and respiratory functions in the nucleus tractus solitarii.