Two highly selective mu-opioid receptor (MOP-R) agonists, endomorphin-1 (EM-1) and endomorphin-2 (EM-2), have been identified and postulated to be endogenous ligands for MOP-R. Experiments were designed to determine the involvement of subtypes of MOP-R on the antinociceptive effects of EM-1 or EM-2 using the paw withdrawal test. The intrathecal (i.t.) injection of EM-1 and EM-2 produced dose-dependent antinociception in mice 1 min after the injection. Subcutaneous (s.c.) pretreatment with naloxonazine (NLZ), a selective MOP1-R antagonist, dose-dependently antagonized the antinociceptive effect of EMs. The antinociceptive effect of EM-2 was more sensitive to NLZ than that of EM-1. The selective heroin/morphine-6beta-glucuronide antagonist 3-methoxynaltrexone (3-MNT) blocked EM-2-induced antinociception, but not EM-1-induced antinociception. The dose-response curve of EM-2 was shifted threefold to the right by pretreatment with s.c. 3-MNT at a dosage of 0.25 mg/kg. EM-2-induced antinociception was attenuated by pretreatment with s.c. nor-binaltorphimine and naltrindole, whereas the effect of EM-1 was not affected. Moreover, the antinociceptive effect of EM-2 was attenuated by i.t. pretreatment with antisera against dynorphin A(1-17) or methionine-enkephalin. These results suggest that EM-2-induced antinociception may be mediated by the subtype of MOP-R, which is sensitive to NLZ and 3-MNT, and by subsequent release of dynorphin A(1-17) and methionine-enkephalin in the spinal cord.