Intracochlear infusion of (+/-)-alpha-amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid (AMPA) was performed with a syringe pump in guinea pigs, and peripheral vestibular dysfunction was induced. Animals were administered edaravone systemically or topically. In the systemic application group, animals were administered edaravone once a day for 7 days after AMPA infusion. In the topical application group, edaravone-soaked gelfoam was placed on the round window membrane just after, 12 h after or 24 h after AMPA infusion. Spontaneous nystagmus was observed after AMPA infusion. Immunohistochemistry for 4-hydroxy-2-nonenal (4-HNE), a marker of free radical-induced lipid peroxidation, was performed 24 h after AMPA infusion. In addition, caloric tests were performed to evaluate vestibular function 1 week after AMPA infusion. Animals in both groups showed decreased spontaneous nystagmus, but results were not significant. Animals treated topically with edaravone within 12 h of AMPA infusion showed normal morphology of the ampullar sensory epithelia of the lateral semicircular canals and showed a good response to the caloric tests. 4-HNE immunoreactivity in the sensory epithelia was very low in these animals. In contrast, untreated animals and animals treated with edaravone systemically or topically 24 h after AMPA infusion showed morphologic hair cell damage, reduced caloric response and remarkable 4-HNE immunoreactivity in the sensory epithelia. These results indicate that topical application of edaravone within 12 h after damage protects the vestibular periphery from free radical-induced toxicity in response to intracochlear infusion of AMPA.