ObjectivesTo compare the estimation of carotid artery stenosis by computed tomography angiography (CTA) based on cross-sectional area versus the smallest diameter measurement, and test the accuracy of both CTA measurements using color Doppler ultrasonography (CDUS) as a reference method.MethodsFor 113 carotid arteries with stenosis ≥50% we analysed the differences in the estimated stenosis level between both CTA methods and CDUS using the Bland-Altman approach. Further, we calculated sensitivity, specificity and plotted receiver operating characteristic (ROC) curves for both CTA methods.ResultsThe mean difference between CDUS and CTA (area) measurements was −0.4% (p = 0.68); between CDUS and CTA (diameter), 20.7% (p < 0.001); and between CTA (area) and CTA (diameter), 21.2% (p < 0.001). Sensitivity and specificity for the CTA (area) method were 81% and 77%, and for CTA (diameter) were 23% and 100%. The area under the curve (AUC) for CTA (diameter) was 0.62 (0.57, 0.66), and for CTA (area) 0.79 (0.71–0.87). The equality test for the two AUCs was <0.0001.ConclusionsCTA (diameter)-based measurements significantly underestimated the degree of carotid stenosis. We recommend the CTA (area) method because of its higher predictive power for a correct stenosis classification and a lack of significant difference in the estimated stenosis level, compared to CDUS.Main messages• Cross-sectional area measurement considers asymmetric shape of the residual vessel lumen.• CTA (diameter) method on average significantly underestimates the true level of stenosis.• CTA (area) method correctly classifies vessels needing surgical intervention.