The purpose of this study was to elucidate brain areas that mediate cannabinoid-induced antinociception as assessed in the tail-flick test. Intracerebroventricular administration of the prototypical cannabinoid, delta 9-tetrahydrocannabinol, and the potent bicyclic analog, CP-55,940, produced antinociception at ED50 values of 373 and 64 nmol/rat, respectively. Hypothermic and cataleptic effects also were observed after i.c.v. administration of CP-55,940, but not delta 9-tetrahydrocannabinol. In contrast, the endogenous cannabinoid, anandamide, failed to elicit any apparent pharmacological effects. Administration of CP-55,940 into the caudate putamen produced catalepsy, but failed to produce either antinociception or hypothermia. Micro-injection of CP-55,940 into the ventrolateral aspect of the periaqueductal gray (PAG), in the region of the dorsal raphe, produced antinociception (ED50 dose = 28 nmol/rat), catalepsy and hypothermia. CP-56,667, the inactive stereoisomer of CP-55,940, failed to produce any effects when injected into the same site. Additional studies demonstrated that pertussis toxin completely prevented the pharmacological effects of CP-55,940 when both agents were administered into the posterior ventrolateral PAG. In contrast, dibutyryl-cAMP failed to attenuate cannabinoid-induced antinociception. Finally, CP-55,940 administered into either the posterior dorsolateral or the anterior ventrolateral areas of the PAG was without effect. These results indicate that the antinociceptive and cataleptic effects of cannabinoids in the PAG are dose-related, exhibit regional specificity and are enantioselective. Moreover, the complete prevention of these pharmacological effects by pertussis toxin pretreatment in the PAG is consistent with the involvement of G proteins. These findings suggest that the posterior ventrolateral PAG may be an important brain area for the antinociceptive and cataleptic effects of the cannabinoids.