The operative findings of 23 carotid arteries were compared with pre-operative duplex scans and angiography. Both duplex and angiography had a high degree of accuracy in detecting haemodynamically significant lesions of 50-99% (88 and 92%, respectively). Their accuracy, however, in correctly predicting the grade of stenosis was significantly lower; 30 and 48%, respectively (P < 0.001). They each exhibited an accuracy of 70% and 78%, respectively, in the detection of ulceration. Overall, both duplex and angiography displayed an accuracy of 87% in indicating the proper management course. The combination of both investigations increased this accuracy to 94.6%. In severely stenotic lesions, duplex to rule out occlusion was not reliable. On statistical analysis, there was no difference between duplex and angiography in predicting haemodynamically significant lesions of 50-99%, estimating the grade of stenosis, the detection of ulceration, or indicating the proper management course. In most situations, duplex alone equalled the accuracy of angiography in the pre-operative assessment of patients for carotid surgery. For stenotic lesions of > 90%, however, we recommend supplemental angiography to rule out occlusion.