The aim of this study was to evaluate diagnostic accuracy of contrast-enhanced MRA (CEMRA) compared with digital subtraction angiography (DSA) in studying neck vessels of 48 patients. In three groups of patients, we used three MRA protocols differing for voxel size to assess if intravoxel dephasing effects could modify accuracy of CEMRA. Accuracy and correlation with DSA results were calculated in all patients and separately in the three groups. A qualitative analysis of the likeness between morphology of the stenosis in CEMRA and DSA images was also assessed. In all patients accuracy and agreement with DSA were 96% and k=0.85 in subclavian arteries, 96% and k=0.84 in vertebral artery, 97% and k=0.88 in common carotid arteries, and 94% and k=0.86 in internal carotid arteries. In the three groups accuracy and agreement with DSA did not show any significant difference. Qualitative analysis of CEMRA and DSA images revealed a better agreement in depicting the morphology of stenosis using a smaller voxel size. The CEMRA represents a powerful tool for the non-invasive evaluation of neck vessels. Overestimation trend of CEMRA is confirmed and the reduction of voxel size, decreasing the dephasing intravoxel effect, allows to have a better overlapping of stenosis morphology on CEMRA compared with DSA, but it does not yield diagnostic gain in the stenosis grading.