The number and size of synaptic contacts made by GABA-immunoreactive axonal boutons were estimated in each layer of the primary visual cortex (area Oc1M) of adult rats by using the dissector method. Immunoreactivity for GABA was detected with the postembedding immunogold technique on ultrathin sections. Targets of GABA synaptic contacts were also identified to predict the sites of GABA influence in the rat visual cortex. For the total cortical depth, 82 million out of an overall population of 666 million synaptic contacts per mm3 of tissue (or 1 in 8 contacts, 12%) were GABA. Layer IV averaged 62% more GABA contacts per unit volume than did any other cortical layer. Consequently, these represented a larger proportion (1 in 6, 17%) of the overall population of layer IV synaptic contacts. This higher number of GABA contacts was not due to a greater density of GABA boutons, but to an increased number of contacts made by each layer IV GABA bouton (mean of 1.4 contacts per bouton compared to 1.1 in other cortical layers). The total area occupied by the contacts on an average GABA bouton was similar in all layers; the higher number of contacts per GABA bouton in layer IV being compensated for by their smaller size. This observed constancy in the area of synaptic contacts suggests the presence of one or more regulatory mechanisms maintaining optimal numbers of the different macromolecules forming the synaptic contacts. The increased density of GABA contacts in layer IV compared to other cortical layers was due to their greater number targeting distal regions of the dendritic tree. Since layer IV receives the vast majority of thalamocortical terminals and since these axons preferentially target dendritic spines, the specific arrangement of GABA synaptic contacts in this layer could be designed to exert a precise inhibition near the site of the thalamic input and thus serve as the structural basis for the strong GABA-related hyperpolarization that followed the excitatory response after physiological stimulations of the thalamocortical pathway.