The peripheral analgesic effects of opiates were evaluated in a rat model of inflammation. The experimental design excluded a potential central nervous system site of action for the observed analgesia. After the injection of carrageenan (CARRA) in the plantar surface of both hind paws, an opiate was injected into one paw and saline was injected into the other paw. The inflamed paws injected with the mu-agonist, fentanyl (0.3 micrograms) or the kappa-agonist, ethylketocyclazocine (10 micrograms) were significantly less hyperalgesic (P less than 0.001 and P less than 0.01, respectively) than were the contralateral inflamed paws injected with saline. At these doses, fentanyl and ethylketocyclazocine were devoid of systemic effects. Another mu-agonist, levorphanol (20, 40, 80, or 160 micrograms) and dextrorphan (160 micrograms), its dextrorotatory isomer, were used next to evaluate opioid specificity. Levorphanol produced a dose-related blockade of CARRA-induced hyperalgesia (P less than 0.005). In contrast, 160 micrograms of dextrorphan was inactive. These results demonstrate that local administration of opiates into an inflamed paw produces a dose-related, stereospecific analgesia restricted to the injected area.