OBJECTIVE To validate the use of Doppler ultrasonography and duplex scanning as the only means of preoperative assessment in carotid disease. METHODS A retrospective study performed between January 1980 and December 1989. METHODS Of 597 carotid endarterectomies carried out by one author, 130 procedures were performed on 118 patients (47 women, 71 men, ranging in age from 49 to 85 years) without preoperative cerebral angiography. Justifications for proceeding without angiography were risk of angiography (27 patients), iodine allergy (8 patients), renal insufficiency (5 patients), technical problems (4 patients) and surgeon's preference (74 patients). METHODS Doppler ultrasonography and duplex scanning were the only means of preoperative assessment on 130 occasions. The studies were done by fully trained radiologists. Until 1982 carotid stenosis was assessed with a continuous wave bidirectional Doppler flowmeter and gray-scale ultrasonography in equivocal cases. Thereafter, a real-time spectrum analyser connected on-line to a Doppler flowmeter was used. Duplex scanning was introduced in 1985. METHODS Surgical observations confirmed the preoperative assessment on 124 occasions (96%). RESULTS Six patients had occlusion of the internal carotid artery (ICA); five patients had recent thrombosis that had occurred between the noninvasive assessment and the endarterectomy. Revascularization was successful in all. In one patient, the ICA was chronically occluded. In 63 patients, the mean pressure in the ICA stump and the mean gradient across the stenosis, measured intraoperatively, were 40 +/- 3 mm Hg (95% confidence interval: 34 to 48 mm Hg) and 55 +/- 3 mm Hg (95% confidence interval: 46 to 58 mm Hg), respectively, which confirmed the severity of the stenosis. CONCLUSIONS The authors conclude that in their institution, carotid endarterectomy based on Doppler ultrasonography and duplex scanning only is reliable and safe.