The aim is to investigate the effect of tube voltage and chest wall thickness on image quality, stenosis measurement, and radiation dose in coronary CT angiography (CCTA) in a phantom study. A phantom with tubes in a box at its center and concentric cylindrical plastic chambers of three layers at its periphery was constructed. The concentric cylinders were filled with oil or left empty to simulate different degrees of obesity. Retrospective CT scanning was performed at different kVps and mAs. Image noise, contrast to noise ratio (CNR), stenosis measurement, and radiation dose were obtained. A CNR higher than 10 was considered to be acceptable for clinical practice. Mean image noise was 51.7 at 80 kVp, 31.6 at 100 kVp, and 24.7 at 120 kVp (P < 0.001). A CNR greater than 10 could be achieved with all the images using 80 kVp as well as using 100 or 120 kVp. However, CNRs at 100 and 120 kVp were significantly higher than the CNR at 80 kVp (P < 0.001). There were no significant differences between 100 and 120 kVp. All stenosis measurements were overestimated. Accuracy of stenosis measurement was significantly correlated with CNR (P < 0.05), but not with kVps. Mean doses were 2.07 mSv at 80 kVp, 3.37 mSv at 100 kVp, and 5.17 mSv at 120 kVp (P < 0.001). CNR per radiation dose was highest at 80 kVp, regardless of chest wall thickness. For CCTA, using 80 kVp with high mAs is the best choice, regardless of chest wall thickness, for minimal radiation dose and sufficient image quality.