In a prospective study the accuracy of cross-sectional echocardiography for determination of left ventricular (LV) volume and ejection fraction (EF) was analyzed in 53 patients with coronary artery disease and compared to that of cineangiocardiography (angio). From the apex of the heart phased-array wide-angle (84 degrees) electronic echocardiograms were received in the RAO-equivalent view. Angios were obtained in a 30 degree RAO view. Using Simpson's rule, end-diastolic (EDV) and end-systolic (ESV) LV volumes were calculated and the EF derived. Left ventricular long axis was transected in eight segments, yielding seven diameters. In 50 of the 53 patients cross-sectional echocardiograms could be recorded. The correlation between cineangiocardiography and cross-sectional echocardiography for EDV was highly significant: r = 0.936, y = 0.667x + 27.1, standard error of estimate (syx) +/- 22.2 ml; for ESV: r = 0.970, y = 0.699x + 14.7, syx +/- 14.5 ml; for stroke volume: r = 0.721, y = 0.503x + 11.3, syx +/- 15.3 ml; for EF: r = 0.909, y = 0.740x + 11.3, syx +/- 6.0%. Angio mean long axis was 10.2 +/- 1.2 cm, cross-sectional echocardiographic long axis 8.7 +/- 1.3 cm. Mean LV diameter determined by cineangiocardiography was longer than when determined by cross-sectional echocardiography. The mean difference reached 2.0 cm in the middle of the LV. Our prospective comparative study revealed that LV volumes and EF were underestimated by cross-sectional echocardiography compared to cineangiocardiography because of a methodological systematic error caused by a tangential cut of the heart. In the RAO-equivalent view the "true" long axis was missed. The high corrleation coefficients, however, indicate that the "true" LV volume and EF can be calculated from the given regression equation. Serial measurements should be legitimated.