The purpose of this study was to compare digital subtraction angiography (DSA) and duplex scanning with 2D-Magnetic Resonance Angiography (MRA) to evaluate the accuracy of MRA in determining carotid stenosis. All three methods were applied to 101 carotid arteries in 51 patients. Diameter stenosis of the internal carotid artery was categorized as follows: 0 to 39%, 40% to 59%, 60% to 94%, 95% to 99%, and occlusion (100%). All images were read in blind fashion by different physicians. DSA is still considered the "gold standard" investigation. In 78 arteries the degree of stenosis according to MRA correlated exactly with that of conventional angiography. In the remaining 23, carotid arteries MRA upgraded the stenosis in 13 and downgraded it in 8. The principal problem is the overestimation of the lesion, which was particularly revealing in lesions of more than 60%. Furthermore to date MRA is not able to evaluate the presence of ulceration. For this reason 2D-MRA alone is not a reliable method for evaluating the presence of carotid artery stenosis.