In an effort to eliminate the inherent neurologic morbidity associated with arteriographic investigation, we have increasingly relied upon duplex scans of the extracranial carotid arteries prior to endarterectomy. The percentage of patients undergoing carotid endarterectomy without arteriograms has increased from 5% in 1984 to 69% during 1988-1989. Initially, carotid endarterectomy without arteriography was limited to patients with hemispheric symptoms and relative contraindications. Over the course of the study from 1984-1989, indications for operation were similar for patients having carotid endarterectomy on the basis of duplex scan alone or following arteriography. The perioperative outcome for these patients undergoing duplex scan (n = 255) and arteriography (n = 484) were similar for stroke (2.4%) versus 2.7%, p = NS) and death (0% versus 0.4%, p = NS). Stratification of groups by indication did not show any significant differences in outcome. Duplex scans were sufficiently accurate to replace preoperative arteriograms in identifying significant stenoses at the carotid bifurcation, including asymptomatic disease. Lack of information regarding intracranial arterial occlusive disease did not adversely affect perioperative outcome. Carotid arteriography can be used selectively when duplex scans are technically difficult, when physical examination or scans suggest either inflow (arch) disease or diffuse, distal internal carotid plaque, or when cerebral symptoms are not sufficiently explained by duplex findings.