The formalin test has been proposed as an animal model of pain produced by tissue injury. Although biphasic nociceptive responses to formalin injection have been well documented, low concentrations (0.125 and 0.5%) of formalin injected into the mouse hindpaw produced only the phasic (acute) paw-licking response, lasting the first 5 min after the formalin injection. To explore the involvement of nitric oxide (NO) in the spinal cord and peripheral system during the acute phase of the formalin test, we examined the effect of intrathecal (i.t.) or intraplantar (i.pl.) injection of L-N(G)-nitro arginine methyl ester (L-NAME), a NO synthase inhibitor in mice. Pretreatment with L-NAME (160 nmol), injected i.t., resulted in a significant inhibition of the paw-licking response induced by 0.125 and 0.5% of formalin. L-Arginine (600 mg/kg, i.p.) but not D-arginine (600 mg/kg, i.p.) reversed the antinociceptive effect of L-NAME on the acute nociceptive response induced by low concentrations of formalin. The i.pl. injection of L-NAME (160 nmol) produced a significant decrease of the late (tonic) phase response evoked by 2.0% formalin without affecting the early (acute) phase response. Similar results have been reported in the case of i.t. injected L-NAME as assayed by the 2.0% formalin test. L-NAME (160 nmol), injected into the plantar paw, gave no significant effect on the acute nociceptive response induced by a low concentration of formalin (0.125%). These results suggest that NO in the spinal cord may be involved in not only the late phase response of the formalin (2.0%)-induced paw-licking, but also at least the acute phase response induced by low concentrations (0.125 and 0.5%) of formalin, while peripheral NO has little effect on the early (acute) phase nociceptive response evoked by formalin (0.125--2.0%) injection.