Conventional time-of-flight (TOF) MR angiography (MRA) in carotid artery stenosis relies on flow-related enhancement to produce signal from vascular structures. Intravoxel phase dispersion, due to vortices, causes loss of signal and is the reason for the tendency to overestimate the degree of stenosis. In contrast-enhanced MRA, intravascular signal is mainly dependent on T1 shortening of the blood. We compared first-pass contrast-enhanced MRA (contrast-enhanced 3D gradient echo, ce3D GRE) and contrast-enhanced 2D TOF (ce2D TOF) sequences with an unenhanced 2D TOF in 13 patients with carotid artery stenosis, assessing delineation of the carotid bifurcation, enhancement of veins and grade of stenosis. The contrast-enhanced techniques produced more morphological detail, the ce3D GRE being superior to the ce2D TOF. Four carotid arteries were reclassified into lesser stenosis categories using the ce3D GRE technique. However, seven carotid arteries (27%) were rated as nondiagnostic on the ce3D GRE, mainly due to masking of the carotid bifurcation by veins. The latter can be avoided by decreasing the acquisition time; on our 1.5-T system we could achieve a minimum time of 23 s per 3D GRE. Further reduction of acquisition time would be necessary to incorporate this method into clinical routine, requiring higher-performance gradients, which are not available in many UK hospitals.