High-frequency intravascular ultrasound (30 and 40 MHz) was applied to study 112 human vascular specimens. The ultrasound images were compared with histologic cross-sections. In 44 out of 58 of the histologically classified muscular arteries, a hypoechoic middle layer was seen in the vessel wall, giving it a three-layered appearance. In 10 arteries, fibrous degeneration of the muscular media resulted in a homogeneous appearance of the vessel walls, whereas atherosclerotic plaque precluded the visualization of the arterial media in four of the arteries. A three-layered appearance was seen in seven of nine histologically classified transitional arteries, and a homogeneous arterial wall was seen in two of the nine. None of the 33 elastic arteries, veins, venous bypass, and Goretex conduits showed a hypoechoic medial layer. Histologically proved fibrous intimal thickening was echographically detected in 32 of 48 specimens (67%). It was noted that these intimal lesions were easier to detect with 40 MHz than with 30 MHz transducers. Hypoechoic areas of lipid deposition were detected in 32 of 36 specimens (89%) and could be distinguished from fibrous plaques. Histologically evident calcium deposits were detected with intravascular ultrasound in 35 of 36 specimens (97%). Measurement of plaque area was only possible in cross sections with a three-layered appearance. Quantitative analysis showed a significantly larger lumen area measured from ultrasonic images (26.3 +/- 21.3 mm2) than from histologic cross-sections (21.8 +/- 16.6 mm2, p less than 0.001), probably because of tissue shrinkage during processing for histology. A significant correlation (r = 0.96, p less than 0.001) between ultrasonic and histologic measurements of lumen areas was observed, with and a negligible interobserver and intraobserver variability. Plaque area and medial thickness correlated well with histology (r = 0.87, p less than 0.001 and r = 0.93, p less than 0.001, respectively). It appears from this in vitro study that intravascular ultrasound is an accurate technique for detection and characterization of atherosclerotic lesions. Vessel lumen area can be measured in most instances, whereas plaque area and medial thickness can only be reliably assessed in muscular arteries in which the hypoechoic media serves as a reference, and shadowing by calcium or attenuation by fibrous plaque components is absent.