Bilateral microinjection of 5 nmol morphine into the posterior hypothalamic area (PHA), periaqueductal gray matter (PAG) or ventral tegmental area (VTA) elicits powerful suppression of nociceptive behaviors in the formalin test, an animal model of injury produced pain. The object of the present study was to determine whether analgesia in the formalin test (50 microl 2.5% formalin injected s.c. in one hindpaw) induced by systemically administered morphine requires opioid action at these sites, or other putative sites of opioid action. Morphine sulphate (6 mg/kg s.c.) produced almost complete analgesia in the second phase of the formalin test (30-50 min after formalin). Bilateral microinjection of the quaternary opioid antagonist naloxone methobromide (NxBr, 28 ng in 0.5 microl, 22 min after morphine) into the PHA completely abolished morphine analgesia, while NxBr into PAG partially reversed analgesia. Microinjection of NxBr into the VTA, central nucleus of the amygdala, habenula, striatum, nucleus accumbens or hypothalamic sites outside the PHA did not antagonize morphine analgesia, although microinjections into some of these sites appeared to reduce the cataleptogenic effects of morphine. The data indicate that the PHA and PAG are probably the primary sites of action of morphine in the formalin test.