Background Choosing a suitable cardiac cycle to measure cardiac chamber dimensions and wall thickness can be a more accurate assessment of cardiovascular disease. Methods Cardiac CT was performed on 137 patients for suspected coronary disease. The parameters of left atrium (LA), left ventricle (LV), right atrium (RA), and right ventricle (RV), as well as the wall thickness of LV were measured in different cardiac phases. The general linear mixed model was used to analyze differences in different phases and the correlation between these parameters and traditional risk factors. ROC analysis was performed to estimate LA enlargement. Results The dimensions of LA, RA, and LV wall thickness achieved the maximum at the phase of 35–45%, and those of LV and RV, at 95–5%. Whereas, the changes of LA-B (antero-posterior diameter), LV-D1 (basal dimension), RA-B (minor dimension) and RV-D2 (mid cavity dimension) were relatively more stable during the cardiac cycle. The maximum LA-B diameter(95%CI 36.92,38.48mm), LV-D1 diameter(95%CI 44.36,45.83mm), RA-B diameter(95%CI 48.75,50.61mm), and RV-D2 diameter(95%CI 30.83,32.84mm) and the maximum interventricular septum thickness( 95%CI 10.79,11.51mm) was acquired. Heart rate (HR) and smoking were potential indicators of LVD2 (mid cavity dimension), while HR and LV myocardial mass were potential indicators of LVD3 (apical-basal dimension). In phase 45%, the cut-off value of LA-B was 37.12mm has high sensitivity of 90.9% for predicting LA enlargement. Conclusion Cardiac chamber dimensions and wall thickness vary with the cardiac phase. Choosing the adaptive cardiac phase for evaluating these parameters obtained by cardiac CT could provide a more accurate clinical measurement. Trial registration retrospectively registered.