Rationale and Objectives. This study compared the image quality obtained with different scanning and reconstruction parameters for electron-beam computed tomographic (CT) angiography and sought optimal methods for visualizing the coronary artery lumen. Materials and Methods. Electron-beam CT angiography with contrast material enhancement was used to image 35 branches of fresh postmortem swine coronary arteries. Different collimation widths, fields of view (FOVs), reconstruction kernels, and algorithms were employed to reconstruct the acquired raw data into CT angiographic images. Image quality was compared and analyzed. Results. The contrast-to-noise ratios (C/Ns) for 1.5-, 2-, and 3-mm section thickness were 28.4 +/- 15.2, 31.9 +/- 9.3, and 33.8 +/- 14.5, respectively (P < .05). The lengths of visualized coronary artery lumina were significantly longer for 1.5-mm scanning (71.6 mm +/- 4.3) than for 2-mm (58.3 mm +/- 5.5) and 3-mm scanning (59.0 mm +/- 8.0) (P < .01). The C/Ns for 12.7-, 18.0-, and 26.0-cm FOV reconstruction were 32.8 +/- 9.9, 28.9 +/- 8.2, and 27.1 +/- 8.2, respectively (not significant), and the visualized luminal lengths were 76.1 mm +/- 12.5, 71.7 mm +/- 14.6, and 65.4 mm +/- 13.1, respectively (not significant). The highest C/N (48.2 +/- 13.3) was achieved with smooth kernels and a cone-beam algorithm, and the lowest (14.7 +/- 3.4) with very sharp kernels and a normal algorithm. Cone-beam algorithm images had significantly higher C/Ns than did normal algorithm images (P < .001), and they demonstrated longer coronary artery lumina (P < .01). Conclusion. Collimation width, FOV, reconstruction kernels, and algorithms are important in the processing of high-quality electron-beam coronary angiograms. A 1.5-mm collimation width, 12.7-cm FOV, cone-beam reconstruction algorithm, and very sharp kernels should help in obtaining the best image quality and depicting the longest segments of coronary artery lumen.