The aim of our study was to establish colour Doppler-assisted duplex imaging (CDDI)-criteria to predict an angiographic internal carotid artery (ICA) stenosis of at least 70%, according to the North American Symptomatic Carotid Endarterectomy Trial (NASCET) and European Carotid Surgery Trialists (ECST) methods of measurement. In the following, we describe the findings in 79 patients who were screened for carotid endarterectomy by CDDI and further evaluated by digital subtraction angiography (DSA). In 158 carotid arteries, 107 stenoses of > 30% and nine occlusions were found. Receiver operator characteristic graphs were constructed in order to calculate sensitivities and specificities of the assessment by CDDI in the prediction of high-grade stenoses determined by angiography. Optimal cut-off points were defined by highest accuracy which reflects the combination of high sensitivity and specificity. The critical index of a high-grade ICA stenosis according to the ECST method could be predicted with an accuracy greater than 90% by a systolic peak velocitiy of 1.25 m/s or an area reduction by CDDI of 70%. Corresponding values, 1.6 m/s and 80% area reduction, predicted the stenosis indexes according to the NASCET method less reliably, with accuracies of between 80% and 90%. Flow velocity criterion was slightly less accurate than the area reduction criterion by CDDI. Finally, double-blind evaluation performed by two readers per examination modality showed that the measurement of area reduction in CDDI is at least as reliable as stenosis indexes according to ECST and NASCET methods.