Background and Purpose-A favorable risk-benefit ratio for warfarin compared with aspirin has been reported for the prevention of major vascular events in symptomatic Ն50% intracranial stenoses. Transcranial color-coded duplex sonography (TCCS) criteria providing an accurate detection of Ն50% and Ͻ50% stenoses of the anterior, middle, and posterior cerebral arteries and basilar and vertebral arteries were evaluated retrospectively with angiography used as the standard of reference. Methods-Prospectively collected TCCS, extracranial color-coded duplex sonography, and intra-arterial digital subtraction angiography data of 310 patients were reviewed. The patients had angiography for confirmation of symptomatic extracranial Ն70% carotid stenoses, symptomatic stenoses (peak systolic velocity higher than the corresponding mean value ϩ2 SDs of 104 normal subjects), and occlusions of the middle cerebral or basilar artery previously assessed by ultrasound. The sonographer was not aware of angiographic findings. Results-TCCS would have detected all 31 of Ն50% intracranial stenoses with 1 false-positive and 35 of 38 Ͻ50% stenoses with 3 false-positives. One of 69 stenoses (1%) and 280 of 2741 normal arteries (10%) were missed because of inadequate insonation windows. The corresponding peak systolic velocity cutoffs for Ն50%/Ͻ50% stenoses were Ն155/Ն120 cm/s (anterior cerebral artery), Ն220/Ն155 cm/s (middle cerebral artery), Ն145/Ն100 cm/s (posterior cerebral artery), Ն140/Ն100 cm/s (basilar artery), and Ն120/Ն90 cm/s (vertebral artery). Conclusions-TCCS may reliably assess Ն50% and Ͻ50% basal cerebral artery narrowing and prove useful for noninvasive management of patients with symptomatic intracranial stenoses. (Stroke. 1999;30:87-92.)