The antagonism of the antinociceptive effects of various kappa-opioid agonists has been studied in the mouse abdominal constriction test. Naloxone produced a much smaller degree of antagonism of U50488H than it did of two other kappa-agonists, U69593 and tifluadom. The kappa-selective antagonist, norbinaltorphimine, also failed to shift the dose-response curve to U50488H in this test, despite producing considerable antagonism of the U50488H effect in the rotarod test and of U69593 in both experimental situations. These results are suggestive of a non-opioid component to the action of U50488H in the abdominal constriction test. At high concentrations, U50488H, but not U69593, also showed non-opioid effects in reducing contractile activity in the field-stimulated isolated guinea-pig ileum, as demonstrated by the profile of antagonism seen with beta-chlornaltrexamine and naloxone. These results suggest that U69593, rather than U50488H, may be the kappa-agonist of choice to use, particularly in in-vivo experiments.