A study in vitro of specimens of human aortic and common carotid arteries was carried out to determine the feasibility of direct measurement (i.e., not from residual lumen) of arterial wall thickness with B mode real-time imaging. Measurements in vivo by the same technique were also obtained from common carotid arteries of 10 young normal male subjects. Aortic samples were classified as class A (relatively normal) or class B (with one or more atherosclerotic plaques). In all class A and 85% of class B arterial samples a characteristic B mode image composed of two parallel echogenic lines separated by a hypoechoic space was found. The distance between the two lines (B mode image of intimal + medial thickness) was measured and correlated with the thickness of different combinations of tunicae evaluated by gross and microscopic examination. On the basis of these findings and the results of dissection experiments on the intima and adventitia we concluded that results of B mode imaging of intimal + medial thickness did not differ significantly from the intimal + medial thickness measured on pathologic examination. With respect to the accuracy of measurements obtained by B mode imaging as compared with pathologic findings, we found an error of less than 20% for measurements in 77% of normal and pathologic aortic walls. In addition, no significant difference was found between B mode-determined intimal + medial thickness in the common carotid arteries evaluated in vitro and that determined by this method in vivo in young subjects, indicating that B mode imaging represents a useful approach for the measurement of intimal + medial thickness of human arteries in vivo.