A directional Doppler ultrasound cerebrovascular examination was compared with angiographical findings of 152 internal carotid arteries. The Doppler examination was abnormal in 36 of 38 (95%) arteries with occlusion or stenosis greater than 75%. Of 63 arteries with lesser degrees of stenosis, the Doppler examination identified only four. There were no false-positive Doppler examinations. If the decision to perform angiography had been predicated exclusively on the presence of abnormal Doppler findings, 61 of 101 (60%) carotid lesions of potential clinical significance would have been overlooked. While the Doppler ultrasound cerebrovascular examination is the most useful noninvasive technique available for the evaluation of certain specific categories of patients with cerebrovascular disease, the technique is based on hemodynamic alterations of pressure and flow, and cannot be expected to identify the relatively large number of non-hemodynamically significant carotid lesions that are still clinically significant as sources of emboli. This paper illustrates that in the routine evaluation of patients with symptomatic cerebrovascular disease, the Doppler examination should not play a part in the decision to proceed with angiography.