The development of a visual prosthesis requires long-term safety evaluation. Functional evaluation using evoked potential, such as visual evoked potential (VEP) and electrical evoked potential, is effective for this safety evaluation. We have experienced fatal cases, likely due to repeated isoflurane (IF) anesthesia, in chronic VEP recording. IF anesthesia may cause malignant hyperthermia, respiratory depression, and even death as a result of repetitive use. Sevoflurane (SF) is a safer anesthetic with a low blood/gas partition coefficient, and the introduction of and arousal from anesthesia can be performed quickly. Since there is no report on VEP in rabbits using SF anesthesia, we investigated whether SF anesthesia can be used as an alternative to IF anesthesia in evoked-potential measurement for the evaluation of the longterm safety of visual prostheses. The first negative wave (N1) was used for the VEP evaluation of rabbits. In both anesthesia methods, obvious N1 peaks were obtained and the latencies were similar. Furthermore, the N1 amplitude under SF was higher than under IF. The difference in N1 amplitudes and latencies at different concentrations of each anesthetic was not statistically significant. Therefore, we found SF to be an alternative to IF in the VEP evaluation of visual prostheses in rabbits.