The ability of high-resolution ultrasound, angiography, and pathologic examination of endarterectomy specimens to identify and quantitate atherosclerosis was compared in a five-center study. The carotid bifurcation in 900 patients was evaluated by angiography and ultrasound. In 216 cases, high-quality endarterectomy specimens were available for comparison with the preoperative images. All comparisons were made in a blinded fashion. Results indicate that ultrasound is able to differentiate angiographically normal from abnormal arteries with a sensitivity of 88% (1077 of 1233 arteries) and accuracy of 79% (1251 of 1578 arteries). Angiographic stenoses equal to or greater than 50% diameter were accurately identified by ultrasound imaging in 72% (1133 of 1578 arteries) of cases, and this was improved by the addition of other functional data (i.e., Doppler spectral analysis and oculoplethysmography). There was only modest correlation of absolute measurements of lesion width, minimal lumen, and standard lumen by the two imaging techniques (r = 0.28 to 0.55). Ultrasound measurements of lesion width were on the average 2 mm greater than those of angiography. The lumen averaged 1.5 mm larger when measured by ultrasound techniques. In the subset in which data were available from endarterectomy specimens, ultrasound showed the best correlation with lesion width (mean difference -1.1 mm) and angiography correlated best with minimal lumen (mean difference -0.1 mm). Neither examination consistently identified ulcerated plaques. Although ultrasound imaging alone has limited usefulness in quantitating luminal stenosis, this can be improved by the use of Doppler spectral analysis and oculoplethysmography. Ultrasound is superior to angiography for quantifying atherosclerotic plaque (lesion width) and will be an important tool for further study of atherosclerotic lesions.