In an attempt to prevent morbidity and mortality in carotid endarterectomy, we monitored 47 operations with intraoperative EEG to determine which patients should have a shunt during endarterectomy. The EEG was recorded for up to five minutes during test carotid cross-clamping. When EEG asymmetry between the cerebral hemispheres occurred, the clamp was immediately removed. A shunt was used in all patients who had EEG asymmetry. In the absence of EEG asymmetry, no shunt was used. Of the 38 patients with no EEG asymmetry intraoperatively, one patient had transient deficits postoperatively. Electroencephalographic asymmetry occurred in nine patients during test carotid cross-clamping. Postoperatively, five of these shunted patients awakened neurologically intact and four patients had transient deficit. When the EEG was normal and no shunt was used, patients did well; when the EEG was abnormal and a shunt was used, there was a high incidence of transient neurologic deficit.