Opiates and opioid peptides were administered in the order of 10(-9)-10(-6) mol peripherally, and their action on pain sensitivity was investigated by the modified formalin test which has two characteristic pain responses (the first and the second phase) in the mouse hindpaw. Opioid peptides (20-500 pmol) had dose-dependent analgesia against both first and second phases, and their action ranked dynorphin greater than [D-Ala2, Met5]-enkephalinamide greater than [Met5]-enkephalin. EKC and morphine (0.4-2.5 nmol) inhibited pain response of the first phase, but produced hyperalgesia in the second phase dose-dependently. Lidocaine hydrochloride had peripheral analgesic action, but was about 500-10000 times weaker than these substances. So, these peripheral analgesic actions have a different mechanism from that of local anesthetic action. N-methyl levallorphan which is thought to be a peripherally selective narcotic antagonist reversed these peripheral analgesic actions at the first and second phases and also prevented the hyperalgesic effects of EKC and morphine at the second phase. Naloxone reversed analgesia at only the first phase. These results suggest that an analgesic mechanism by opioids may exist at the peripheral site as well. Furthermore, it is estimated that a receptor exists which is antagonized by N-methyl levallorphan but not by naloxone and that there is a system of hyperalgesia by EKC and morphine in pain modulation.