The effects of buprenorphine (BNP, 10-200 μg/kg, i.v.) and pentazocine (PTZ, 2.5-10 mg/kg, i.v.) on the regional cerebral metabolic rate for glucose (rCMR_glc) were analyzed in nine anatomically discrete areas of the conscious rat brain by the simultaneous use of [¹⁴C]2-deoxyglucose, the glucose analogue that can be phosphorylated in the brain, and [³H]3-O-methylglucose, a nonmetabolizable glucose analogue. Originally, this method was developed by Gjedde and Diemer in the rat and in humans. The rCMR_glc was significantly decreased by BNP (100 or 200 μg/kg) in most of the brain regions investigated, except the cerebellum. In contrast, PTZ (10 mg/kg) significantly increased rCMR_glc in the cerebral cortex and medulla. In the cerebral cortex and medulla, the direction of the effect on rCMR_glc was opposite for BNP (22% decrease at the dose of 200 μg/kg) and PTZ (22% increase at the dose of 10 mg/kg). These findings strongly suggest that the discrepancies between the marked effects of BNP (a partial μ agonist and κ antagonist) and PTZ (a μ antagonist and κ agonist) on rCMR_glc reflect the selectivity of agonist action at the different types of opioid receptors, μ and κ receptors, in the rat brain.