Conflicting reports on the usefulness of intraoperative monitoring of visual function by means of visual evoked potentials (VEPs) initiated this study. In 32 patients without visual problems, VEPs were recorded to evaluate the reliability for intraoperative monitoring with total intravenous anesthesia. All patients underwent noncranial surgery. Using a standard technique, VEPs were recorded preoperatively in the awake patients and after induction of anesthesia during surgery. A total of 1436 intraoperative traces were recorded and analyzed. A minor prolongation of the P100 latency of 8% and a more pronounced attenuation of the P100-N145 amplitude of 60% were observed in the anesthetized patients. In most of the anesthetized patients, a stable recording of VEPs was not obtainable. In 4 patients (12.5%), clearly identifiable VEP peaks were detected in more than 90% of the traces recorded intraoperatively. In 88% of the patients, reproducible VEPs were obtained in less than 75% of the intraoperative traces only. We concluded that with standard recording techniques and total intravenous anesthesia, intraoperative VEP monitoring in surgically anesthetized patients is not reliable.