The authors evaluated the accuracy of "black blood" magnetic resonance (MR) angiography for depicting disease involving the extracranial carotid arteries. Two- and three-dimensional flow-compensated gradient-echo sequences were employed to create "bright blood" images. A thin-section spin-echo sequence with flow presaturation allowed the creation of black blood images. Projection angiograms were made from bright and black blood images with application of a maximum- or minimum-intensity projection algorithm, respectively. These methods were used in 13 healthy volunteers and 17 patients, and a prospective blinded comparison of MR angiography and conventional angiography was performed. Normal carotid arteries were well shown with both bright and black blood methods; in patients, both methods were sensitive for detecting carotid disease. However, bright blood angiography exaggerated the severity of carotid lesions in 13 of 33 arteries, mostly in severe disease; this problem was not encountered with black blood angiography. The authors conclude that bright blood angiography is a sensitive method for screening carotid disease; when a significant abnormality is found, black blood angiography should be performed for more precise delineation of the lesion.