The ability to classify lesion composition accurately may be important for selecting or guiding interventional therapy and for understanding the pathophysiologic basis of individual lesions. To assess the usefulness of ultrasound in classifying lesions, intracoronary ultrasound images were obtained from 44 atherosclerotic lesions in patients before directional atherectomy. Lesions were classified by visual analysis and by computer-assisted gray-level statistics. Atherectomy samples were evaluated histologically for elastosis and calcium and quantitatively by morphometric analysis for various tissue components. The computer-assisted quantitative classification agreed well with the findings on visual analysis. Visual and computer-assisted quantitative ultrasound images were found to have distinctive histologic features. Lesions with predominantly echogenic plaque had a larger fraction of dense fibrous, elastic, or calcified tissue. Lesions with predominantly echolucent soft plaque had a greater fraction of loose fibrous, smooth-muscle, thrombotic, or necrotic elements. Thus intracoronary ultrasound allows accurate classification of lesion composition in patients.