Objectives This study aimed to investigate the influence of vessel length on transluminal attenuation gradient (TAG) and establish a new index, VLN-TAG (VLN-TAG (HU/100 mm(2))=TAG (HU/10 mm)/vessel length (10 mm)), to estimate the diagnostic value using 320-slice computed tomography (CT). Methods A total of 150 coronary arteries from 52 patients who underwent single-beat scanning using 320-slice CT and invasive coronary angiography (ICA) within 2 weeks were retrospectively enrolled. TAG was obtained from the major three epicardial vessels, and its interrelation with the measured length of the vessels was evaluated by Pearson correlation and regression analyses. The changes in TAG and VLN-TAG were compared with the corresponding stenosis severities ascertained by ICA using repeated measures ANOVA. Results TAG had a significant interrelation with the measured length of the vessels (r=0.644, p<0.001). Neither TAG nor VLN-TAG with different stenosis degrees of <50, 50-70, and 70-99% on ICA had significant difference among the three groups. Plaque composition had no influence on VLN-TAG in all groups. The combined TAG or VLN-TAG and coronary computed tomography angiography (CCTA) assessment did not significantly change the area under the curve compared with using CCTA only. In the calcified vessels group, adding VLN-TAG to CCTA improved the specificity (92.86 vs 85.71%). Conclusions Vessel length is an important factor impacting TAG. TAG does not offer an incremental diagnostic value compared with CCTA only for detecting coronary stenosis.