Plaque volume determined by common linear 3-D IVUS analysis systems will show under- or overestimation in curved vessel segments because these systems approximate the true 3-D transducer pull-back trajectory by a straight line. We developed a mathematical model that showed that the error is primarily dependent on the curvature of the pull-back trajectory and not on vessel tortuosity. Furthermore, we measured this error in vivo in the coronary arteries of 15 patients, comparing the plaque volume using a true 3-D reconstruction method with that of the linear approach. The in vivo plaque volume error ranged from 2.3% to -1.2% for 15 coronary segments with lengths ranging from 38.8 to 89.1 mm (62.2 +/- 13 mm). The volume error introduced by linear 3-D IVUS analysis systems is dependent on the curvature of the pull-back trajectory. The error measured in vivo was small and inversely related to segment length.