Magnetic resonance (MR) angiography offers a safe, noninvasive alternative to conventional angiography in patients with suspected carotid stenosis; however, it tends to overestimate the severity of stenosis. Loss of the MR signal with a resulting flow gap is a frequent finding in cases of high-grade stenosis. The authors undertook this study to define the range of carotid stenosis associated with a flow gap on two-dimensional time-of-flight (2DTF)-MR angiography. Blinded evaluations were made of 102 common carotid bifurcations in 51 patients who had undergone both conventional angiography and 2DTF-MR angiography. The percent of diameter stenosis was calculated from the conventional angiogram using the method adopted by the Asymptomatic Carotid Atherosclerosis Study (ACAS) trial. An MR flow gap was noted if there was a segment of the vessel that was completely free of signal with a reappearance of the signal distally. According to conventional angiography, the minimum percentage of stenosis associated with a flow gap is 56%. Flow gaps were present in 20 of 22 arteries (sensitivity 91%) with stenosis of 60% or more and in two of the 66 arteries (specificity 97%) with less than 60% stenosis. Flow gaps were present in all arteries with stenosis of 70% or more. Complete occlusion was correctly identified in 10 of 10 cases. These results demonstrate that the presence of a flow gap on 2DTF-MR angiography is a reliable marker of clinically significant carotid stenosis (measuring 60% or more), with sensitivity and specificity comparable to duplex carotid ultrasound. In addition, MR angiography can be used to screen the intracranial circulation for significant vascular pathology in patients being considered for carotid endarterectomy.