Putative GABAergic elements in the pontine nuclei have been studied in the rat, cat and two old world monkeys (Macaca mulatta and Papio papio) using an antiserum against GABA-glutaraldehyde-protein conjugates and the peroxidase-antiperoxidase method. In addition, an antiserum against glutamate decarboxylase has been used in the cat. For comparison, Golgi impregnated material from cat and macaque has been studied. In all species there is a moderately dense plexus of fibres with GABA-like immunoreactivity with only minor regional differences between different parts of the pontine nuclei. The number of cell bodies showing GABA-like immunoreactivity is, however, strikingly different. Thus, in the rat there are very few such neurons. In the cat, they make up about 1% of the total cell population, while the corresponding number in the two primate species is about 5%. The number is consistently somewhat higher in rostral than in caudal parts of the pontine nuclei. Numbers in the cat are essentially the same with the glutamate decarboxylase antiserum as with the GABA antiserum. The size of GABA-like immunoreactivity positive somata is very similar in cat, macaque and baboon, averaging about 160 micron2 in cross-sectional area. The average cross-sectional area of the total neuronal population as measured in adjacent thionin-stained sections is about 280 micron2. However, the range of sizes for GABA-like immunoreactivity positive cells is wide, so that size alone is not a good criterion for their identification. Although their dendritic morphology is varied, a significant proportion of GABA-like immunoreactivity positive cells have very long and straight dendrites. A few examples were found in the primate species of GABA-like immunoreactivity positive cells with processes tentatively identified as axons. Such processes could be seen to divide several times. No such branching processes could be identified, however, in Golgi impregnated material from the same species. In order to determine whether GABA-like immunoreactivity positive cells project to the cerebellum, retrograde tracing of horseradish peroxidase-wheat germ agglutinin was combined with immunocytochemistry. No double labelled cells could be found in the pontine nuclei. Comparison of size distribution of retrogradely labelled pontocerebellar and GABA-like immunoreactivity positive cell bodies showed a high degree of overlap, although the average size of projection neurons and GABA-like immunoreactivity positive ones is clearly different.(ABSTRACT TRUNCATED AT 400 WORDS)