One hundred carotid endarterectomies were performed using selective shunting based on continuous electroencephalographic monitoring (CEM) for the detection of cortical ischemia. Changes associated with ischemia were loss of frequency and amplitude. The results of CEM were correlated with carotid stump pressure (CSP) measurements. Only one (4%) of the 25 patients who developed an abnormal EEG had a mean CSP greater than 50 mmHg: however, CEM was positive in only 24 (45%) of the 53 patients with a CSP less than 50 mmHg. Using EEG as a standard, no CSP criterion (50 or 25 mmHg) was sufficiently sensitive and specific to recommend its routine use (50 mmHg--96% and 61% respectively; 25 mmHg--40% and 96% respectively). Operative mortality was 1%. The three intraoperative neurologic deficits (3%), one transient and two permanent, are analyzed with respect to the operative EEG findings. In the 100 endarterectomy patients and three more undergoing carotid exploration or excision of carotid body tumor EEG changes due to anesthetic problems not associated with carotid clamping were often identified. In 48 additional endarterectomies a computerized display and disk storage of six selected EEG leads, each with a reference trace for comparison, has provided information equivalent to that from the full EEG. EEG monitoring is more accurate than CSP measurement for identifying patients who require shunting, safely dispensing with a shunt in many cases which would otherwise be shunted by standard CSP criteria.(ABSTRACT TRUNCATED AT 250 WORDS)