The accuracy of catheter-based intravascular ultrasonography to define luminal size in humans in vivo and its sensitivity to describe lesion morphology have not been previously reported. Vessel diameter, cross-sectional area and lesion characteristics assessed by digital subtraction angiography and intravascular ultrasonography (20 MHz) were compared in 86 human arterial segments. The same arterial segments were imaged and analyzed by digital subtraction angiography and intravascular ultrasonography at 49 femoral, 3 renal, 5 iliac, 7 pulmonary and 22 aortic sites. Digital subtraction angiographic diameter and area were determined geometrically by an automated algorithm. Intravascular ultrasonographic diameter and area were determined by planimetry. Linear correlation for diameter by the two techniques was 0.97, standard error of the estimate (SEE) = 1.83 mm, and for cross-sectional area it was 0.95, SEE = 0.65 cm2. Intravascular ultrasonography identified 24 sites in which plaque was present; 11 (46%) of these segments appeared normal by digital subtraction angiography. Conversely, digital subtraction angiography demonstrated irregularities in 18 segments of which 5 (28%) appeared normal by intravascular ultrasonography. These data indicate an excellent correlation between intravascular ultrasonography and digital subtraction angiography for in vivo assessment of human arterial dimensions in normal and minimally diseased segments. However, intravascular ultrasonography is more likely to identify atherosclerotic plaque that may be angiographically "silent."