Purpose: The objectives of the study were to evaluate the validity of intraoperative flash visual evoked potentials (VEPs) when electroretinograms (ERGs) were unstable, to compare white versus red light-emitting diodes, and to assess the impact of luminance on ERG variability. Methods: Thirty patients were included (Inomed system; pre- and postoperative visual fields). Possible changes in visual fields were assessed with mean defects in perimetry. The receiver operating characteristic (ROC) curves of normalized VEPs and of normalized and corrected VEPs with ERGs were compared. Results: Thirty-two eyes could be analyzed in 20 patients (mainly gliomas and meningiomas): 2 had a severe defect in their visual field, and 6 had a mild defect. The receiver operating characteristic curve indicated (1) normalized and corrected VEPs with ERGs were more reliable than normalized VEPs only (P < 0.03) and (2) an alarm threshold of 80% of normalized and corrected VEPs. No significant difference in variability of ERGs was found with white or red light-emitting diodes with this system. Increased luminance improved stability of ERGs (P < 0.05). Conclusions: Normalization and correction of VEPs with ERGs improved the validity of VEPs and indicated a 20% decrease as alarm criterion. This normalization and correction with peripheral excitation could be generalized to improve the reliability of neuromonitoring.