Endocannabinoids, acting via type 1 cannabinoid receptors (CB1), are known to be involved in short-term synaptic plasticity via retrograde signaling. Strong depolarization of the postsynaptic neurons is followed by the endocannabinoid-mediated activation of presynaptic CB1 receptors, which suppresses GABA and/or glutamate release. This phenomenon is termed depolarization-induced suppression of inhibition (DSI) or excitation (DSE), respectively. Although both phenomena have been reported to be present in the basal ganglia, the anatomical substrate for these actions has not been clearly identified. Here we investigate the high-resolution subcellular localization of CB1 receptors in the nucleus accumbens, striatum, globus pallidus and substantia nigra, as well as in the internal capsule, where the striato-nigral and pallido-nigral pathways are located. In all examined nuclei of the basal ganglia, we found that CB1 receptors were located on the membrane of axon terminals and preterminal axons. Electron microscopic examination revealed that the majority of these axon terminals were GABAergic, giving rise to mostly symmetrical synapses. Interestingly, preterminal axons showed far more intense staining for CB1, especially in the globus pallidus and substantia nigra, whereas their terminals were only faintly stained. Non-varicose, thin unmyelinated fibers in the internal capsule also showed strong CB1-labeling, and were embedded in bundles of myelinated CB1-negative axons. The majority of CB1 receptors labeled by immunogold particles were located in the axonal plasma membrane (92.3%), apparently capable of signaling cannabinoid actions. CB1 receptors in this location cannot directly modulate transmitter release, because the release sites are several hundred micrometers away. Interestingly, both the CB1 agonist, WIN55,212-2, as well as its antagonist, AM251, were able to block action potential generation, but via a CB1 independent mechanism, since the effects remained intact in CB1 knockout animals. Thus, our electrophysiological data suggest that these receptors are unable to influence action potential propagation, thus they may not be functional at these sites, but are likely being transported to the terminal fields. The present data are consistent with a role of endocannabinoids in the control of GABA, but not glutamate, release in the basal ganglia via presynaptic CB1 receptors, but also call the attention to possible non-CB1-mediated effects of widely used cannabinoid ligands on action potential generation.