We compared duplex scanning, "bright blood" and "black blood" magnetic resonance angiography, and conventional angiography for evaluation of the extracranial carotid arteries. All three methods were applied to 39 vessels in 20 patients. Duplex scanning was inaccurate when compared to conventional angiography in six instances. In three instances the degree of stenosis was overgraded by the scanner, and in three cases the stenosis was undergraded. Magnetic resonance angiography was inaccurate when compared to conventional angiography in three instances. In all cases magnetic resonance angiography overgraded the degree of stenosis. By use of a greater than 70% stenosis as a positive study, the sensitivity of magnetic resonance angiography was 100% and the specificity 92%. With use of the same criteria, the sensitivity of duplex scanning was 86%, and specificity was 84%. In those evaluations where the results of the magnetic resonance angiography and duplex scanning were in agreement, the correlation with conventional angiography was 100%. We conclude that magnetic resonance angiography is an alternative means to duplex scanning for noninvasive carotid imaging. A combination of bright and black blood magnetic resonance angiography is precise in delineating lesions of the extracranial carotid artery and may ultimately eliminate the need for conventional angiography in the evaluation of carotid stenosis.