Controversy continues concerning the advisability of routine shunting, no shunting, or selective shunting during carotid endarterectomy. Because of its reflection of the physiologic state of the end organ, the authors chose routine 18 lead EEG monitoring as a guide to selective shunting and as an indication of adequate shunt function during all carotid endarterectomies performed from December 1977 through July 1982. In that period, 200 patients underwent 219 endarterectomies under general anesthesia and EEG monitoring. Ischemic EEG changes at the time of carotid cross clamping suggested the need for intraluminal shunts in 16% of patients. Insertion of shunts restored the EEG pattern to normal in all instances, although in two patients, adjustment of the shunt was required to maintain this results. EEG changes requiring shunting occurred in 10% of patients with unilateral disease, in 27% of patients with bilateral disease, and in 42% of patients with unilateral stenosis and contralateral occlusion. Twenty-seven patients had small fixed neurologic deficits before operation. Surgery was not delayed in these individuals who demonstrated no increased requirement for shunts and no new postoperative neurologic deficits. In the group of 150 endarterectomies performed as separate procedures, there was one (0.7%) fixed neurologic deficit after operation, one transient deficit (0.7%), and one death (0.7%). Sixty-nine endarterectomies were performed simultaneously with open heart surgery and were associated with one fixed neurologic deficit (1.4%) and two transient deficits (2.9%). All four deaths in this group were attributable to the cardiac surgical procedures. These results indicate that selective shunting based on EEG monitoring permits the safe performance of carotid endarterectomy, even in patients considered to be at high risk for postoperative neurologic deficit.