OBJECTIVE. The sonographic diagnostic criteria for vertebral artery stenosis have not been fully investigated. The purpose of this study was to assess hemodynamic parameters at color Doppler imaging and to determine, with digital subtraction angiography as the reference standard, the optimal thresholds for evaluation of proximal vertebral artery stenosis. MATERIALS AND METHODS. Among 653 patients with symptoms of ischemia of the posterior circulation, 247 subjects with normal arteries or stenosis of the proximal vertebral artery confirmed with digital subtraction angiography were included in the study. Peak systolic velocity at the origin of the vertebral artery (PSVorigin) and in intervertebral segments of the vertebral artery (PSVIV), end-diastolic velocity at the origin and in the intervertebral segments of the vertebral artery, and the diameter of the vascular lumen were measured. The cutoff values for the diagnosis of < 50%, 50-69%, and 70-99% stenosis were determined with receiver operating characteristics analysis. RESULTS. The optimal cutoff values of hemodynamic parameters in evaluation of stenosis of the proximal vertebral artery for < 50% stenosis were PSVorigin >= 85 cm/s, PSVorigin /PSVIV >= 1.3, and end-diastolic velocity at the origin >= 27 cm/s; for 50-69% stenosis were PSVorigin >= 140 cm/s, PSVorigin/PSVIV >= 2.1, and end-diastolic velocity at the origin >= 35 cm/s; and for 70-99% stenosis were PSVorigin >= 210 cm/s, PSVorigin/PSVIV >= 4.0, and end-diastolic velocity at the origin >= 50 cm/s. PSVorigin was the most useful hemodynamic parameter, having accuracy of 94.5%, 96.2%, and 88.7% for the diagnosis of < 50%, 50-69%, and 70-99% stenosis. CONCLUSION. Color Doppler imaging is a reliable method for evaluation of vertebral artery stenosis. The results derived from this study can be used as a reference for establishing sonographic criteria for proximal vertebral artery stenosis.