Objective The goal of this study was to determine the performance of intraoperative visual evoked potentials (VEPs) in detecting visual field changes. Methods Assessments of VEPs were performed with simultaneous retinal responses by using white light-emitting diodes protected from scialytic microscope lights. The alarm criterion was a reproducible decrease in amplitude of VEPs wave P100 of 20% or more. Visual fields were assessed preoperatively and 1 month postsurgery (Goldmann perimetry). Results The VEPs were analyzed for 29 patients undergoing the resection of a brain lesion. In 89.7% of patients, steady VEP and retinal responses were obtained for monitoring. The absence of alarm was associated in 94.4% of cases with the absence of postoperative visual changes (specificity). The alarms correctly identified 66.7% of cases with any postoperative changes, and 100% of cases with changes more severe than just a discrete quadrantanopia or deterioration of an existing quadrantanopia (sensitivity, new diffuse deterioration < 2 dB). In 11.5% of patients, a transitory VEP decrease with subsequent recovery was observed without postoperative defects. Conclusions Intraoperative VEPs were performed with simultaneous recording of electroretinograms, with protection from lights of the operating room and with white light-emitting diodes. Intraoperative VEPs were shown to be reliable in predicting postoperative visual field changes. In this series of intraaxial brain procedures, reliable intraoperative VEP monitoring was achieved, allowing at minimum the detection of new quadrantanopia. The standardization of this technique appears to be a valuable effort in regard to the functional risks of homonymous hemianopia.