Because of controversies in the cerebrovascular literature regarding the use of an intraluminal shunt in carotid endarterectomy, we report a randomized prospective study of 118 consecutive symptomatic patients receiving surgery within a single neurosurgical practice. Over 4 years, 138 carotid endarterectomies were performed in the 118 patients, 63 operations with intraluminal shunting and 75 without. Standard rationale for surgery included ipsilateral cerebral infarction in 38% of the operations and ipsilateral transient ischemic attacks in 36%. Unilateral angiographic stenosis of greater than 90% was seen in 58% of the operations; there were no ipsilateral occlusions. Surgery was performed under general anesthesia with barbiturate induction and mild blood pressure elevation. The 30-day complication rate included a mortality rate of 0.7% with a 5.1% incidence of postoperative neurologic deficit and a 1.4% rate of myocardial infarction. In the 24 hours after surgery there were no cerebral infarctions in the shunted group and six in the unshunted group. This 8% rate in the unshunted group compared with 0% in the shunted group was significant at p = 0.023 with a power of 0.95 by Fisher's exact test and chi 2 analysis. This suggests that in our neurosurgical practice (resident training program) the use of an intraluminal shunt during carotid endarterectomy significantly reduces the risk of intraoperative neurologic deficit without increasing the incidence of other complications.