The assumed low ability of the quaternary narcotic antagonist N-allyl levallorphan-bromide (CM 32191) to cross the blood-brain barrier and its selectivity in relieving the peripherally-elicited antipropulsive action of morphine while preserving analgesia has been tested. To ascertain the extent of penetration of CM 32191 into the cns, its relative potency in preventing the in-vivo binding of high specific activity [3H]buprenorphine in the rat cns and small intestine was compared. Pretreatment was with CM 32191 at 16, 30 or 60 mg kg-1 s.c., 20, 60 or 120 min before buprenorphine, the concentrations of which in cerebrum and spinal cord were comparable with control values, but were consistently reduced in the intestine (longitudinal muscle with attached myenteric plexus). Pretreatment with naloxone (20 min, 0.5 or 1 mg kg-1 s.c.) lowered buprenorphine binding in intestine and cns. Neither narcotic antagonist produced significant changes in buprenorphine plasma concentrations. The peripheral selectivity of CM 32191, methyl naloxone and naloxone was examined by investigating in the same rats nociception in the hot plate (central opiate-sensitive mechanism) and the transit of a charcoal meal along the small intestine (local opiate-sensitive mechanism). Both effects were inhibited by morphine (5 mg kg-1 i.v.). Naloxone (10 min pretreatment, 0.5 or 1 mg kg-1 s.c.) did not selectively antagonize intestinal action of the morphine since the relief of charcoal transit inhibition was consistently associated with complete loss of analgesia.(ABSTRACT TRUNCATED AT 250 WORDS)