Without strong evidence of benefit, the use of carotid endarterectomy for prophylaxis against stroke rose dramatically until the mid-1980s, then declined. Our investigation sought to determine whether carotid endarterectomy reduces the risk of stroke among patients with a recent adverse cerebrovascular event and ipsilateral carotid stenosis. We conducted a randomized trial at 50 clinical centers throughout the United States and Canada, in patients in two predetermined strata based on the severity of carotid stenosis--30 to 69 percent and 70 to 99 percent. We report here the results in the 659 patients in the latter stratum, who had had a hemispheric or retinal transient ischemic attack or a nondisabling stroke within the 120 days before entry and had stenosis of 70 to 99 percent in the symptomatic carotid artery. All patients received optimal medical care, including antiplatelet therapy. Those assigned to surgical treatment underwent carotid endarterectomy performed by neurosurgeons or vascular surgeons. All patients were examined by neurologists 1, 3, 6, 9, and 12 months after entry and then every 4 months. End points were assessed by blinded, independent case review. No patient was lost to follow-up. Life-table estimates of the cumulative risk of any ipsilateral stroke at two years were 26 percent in the 331 medical patients and 9 percent in the 328 surgical patients--an absolute risk reduction (+/- SE) 17 +/- 3.5 percent (P less than 0.001). For a major or fatal ipsilateral stroke, the corresponding estimates were 13.1 percent and 2.5 percent--an absolute risk reduction of 10.6 +/- 2.6 percent (P less than 0.001). Carotid endarterectomy was still found to be beneficial when all strokes and deaths were included in the analysis (P less than 0.001). Carotid endarterectomy is highly beneficial to patients with recent hemispheric and retinal transient ischemic attacks or nondisabling strokes and ipsilateral high-grade stenosis (70 to 99 percent) of the internal carotid artery.