Flow velocity measurement by duplex ultrasonography (DUS) is widely accepted as a preoperative method for grading carotid stenoses. Diameter-based grading is used for angiographic methods, but so far the experience is limited regarding diameter-based grading using DUS. In 101 cases with different degrees of carotid stenoses, we compared diameter-and velocity-based grading using DUS and evaluated variability and reproducibility for diameter measurements in 38 of the stenoses. In 18% of the cases, suboptimal image quality with calcifications and acoustic shadowing precluded diameter measurements. In the remaining 83 cases, the agreement between velocity-based and diameter-based grading for distinction between significant and non-significant stenosis was 69% with a kappa value of 0.40. Repeated measurements of diameters showed a coefficient of variation of 10% and intraclass coefficients of correlation of 0.9. We conclude that diameter measurement with DUS can be performed with high reproducibility in most patients with carotid stenosis. These measurements can then be used for calculations of the degree of stenosis, to supplement velocity-based grading. The moderate agreement between diameter-based and flow velocity-based grading cannot easily be explained. Discrepancies probably constitute a mixture of methodological errors and hemodynamic factors and need further evaluation.