Unanesthetized rats, made physically dependent over 5 days by chronic intra-arterial infusion of increasing concentrations of morphine (35-100 mg/kg/day) underwent withdrawal by naloxone (6 micrograms) injection into either the lateral ventricle (i.c.v.), fourth ventricle (V4), intrathecal subarachnoid space (i.t.), or intra-arterially (i.a.) and were evaluated for cardiovascular and behavioral signs of precipitated abstinence. Naloxone i.c.v. produced a significantly greater increase in the magnitude and duration of withdrawal hypertension than did V4 injection. Naloxone i.t. produced a distinctively different, persistent, pressor response as compared to i.c.v., V4 or i.a. routes of administration, although no quantitative differences in behavioral signs of withdrawal were observed. Morphine-dependent, spinal transected (C1) animals generated an augmented pressor response to i.c.v. or i.t. naloxone. This pressor response was accompanied by a significant reduction in core temperature (0.50-0.79 degrees C). Both the naloxone-induced pressor and hypothermic responses were abolished by ganglionic (hexamethonium, 100 mg/kg, i.a.) or peripheral alpha-adrenergic (phentolamine 4 mg/kg, i.a.) blockade. The hypertensive and hypothermic effects of naloxone also were prevented in transected dependent rats by prior spinal pithing. We conclude that in morphine-dependent rats: supraspinal sites rostral to the V4 mediate a more intense naloxone-induced pressor response than caudal regions; cardiovascular and behavioral signs of withdrawal can be precipitated via the spinal cord of intact animals; and the production of withdrawal hypertension and hypothermia in spinal transected morphine-dependent rats indicates that these abstinence signs can be mediated through neuronal pathways within the spinal cord.