A modified formalin test in mice was investigated. The pain response curve induced by 0.5% formalin was biphasic, having 2 peaks, from 0 to 5 min (first phase) and from 15 to 20 min (second phase). A low concentration of formalin was used, allowing the effects of weak analgesics to be detected. Centrally acting drugs such as narcotics inhibited both phases equally. Peripherally acting drugs such as aspirin, oxyphenbutazone, hydrocortisone and dexamethasone only inhibited the second phase. Aminopyrine and mefenamic acid which acted on both central and peripheral sites inhibited both phases, but the second phase was inhibited by lower doses. Thus, this method enables one to easily distinguish the site of action of analgesics. Furthermore, pain response in the first phase was inhibited by capsaicin-treated desensitization and Des-Arg9-(Leu8)-bradykinin (bradykinin inhibitor). The second phase was inhibited by compound 48/80 pretreatment, indomethacin and bradykinin inhibitor. Therefore, it is suggested that substance P and bradykinin participate in the manifestation of the first phase response, and histamine, serotonin, prostaglandin and bradykinin are involved in the second phase. These results indicate that the first and second phase responses induced by formalin have distinct characteristic properties, and it is a very useful method for examining pain, nociception and its modulation by pharmacological or other means.