Abstract Background Coronary computed tomography angiography (CCTA) may be useful for noninvasive follow-up; however, evaluation of coronary stenosis and CT number of plaque may be inaccurate under different vessel enhancement of contrast media. We examined the reproducibility of the CT number of repeat CCTA using our original CT number-controlling system (CTN-CS), which selects contrast level by a multiple regression equation using body surface area and peak CT number and peak time on timing bolus and during CCTA. Methods and results One hundred seventy-two patients who underwent serial CCTA were prospectively and randomly assigned to 3 groups. In the first group, Group A, the amount of contrast for the second CCTA was determined by CTN-CS to match the intracoronary CT number of the first CCTA. In Group B, each patient received the same amount of intravenous contrast in both CCTA examinations. In Group C, 0.7 mL/mg body weight (BW) of contrast medium (350 mgI/mL) was used for baseline and follow-up CCTAs. The regression of repeated CCTAs was the best in Group A ( r = 0.85, p < 0.001) vs. Group B ( r = 0.52, p < 0.001), and Group C ( r = 0.61, p < 0.001). The absolute difference between intracoronary CT numbers of the second and first CCTA was the lowest in Group A (24.8 ± 21.8 HU), followed by Group B (37.6 ± 26.2 HU; p < 0.05) and Group C (46.5 ± 34.4 HU; p < 0.001). Conclusions Using CTN-CS, the difference of intracoronary CT numbers of the second and first CCTA was the smallest when compared to CCTAs using the same contrast volumes or constant volumes per body weight.