Eighty-one patients suspected of having cerebrovascular disease had 157 carotid arterial systems studied by both duplex ultrasonography and contrast arteriography to better define the role of carotid duplex scanning in the surgical decision-making process. These studies were reviewed in a blinded fashion in conjunction with history and physical examination data by two surgeons, one operating on only symptomatic lesions, the other operating on both symptomatic and asymptomatic lesions. Results were analyzed to ascertain if there was agreement regarding decisions for carotid endarterectomy based on scan findings compared with decisions based on arteriographic findings. Scans were also compared with arteriograms and data were analyzed by decision matrix analysis. The accuracy of duplex scanning in relation to arteriography was 81% for detection of disease, 90% for the detection of ulceration, 83% for the detection of a critical stenosis, and 99% for the detection of total arterial occlusion. There was agreement between the two studies regarding the need for carotid surgery in 91% and 89% of carotid arteries, according to surgeons A and B, respectively. Regardless of the surgeons' indications for carotid endarterectomy, duplex ultrasonography provides sufficient information for proper surgical decision making in a high percentage of patients. The accuracy of duplex scanning and the risks of contrast arteriography suggest a possible future role for the routine use of duplex ultrasonography with selective utilization of arteriography in the surgical decision-making process in patients being evaluated for cerebrovascular occlusive disease.