Background and Purpose-Intracranial atherosclerotic disease is associated with a high risk of stroke recurrence. We aimed to determine accuracy of transcranial Doppler screening at laboratories that share the same standardized scanning protocol. Methods-Patients with symptoms of cerebral ischemia were prospectively studied. Stroke Outcomes and Neuroimaging of Intracranial Atherosclerosis (SONIA) criteria were used for identification of Ն50% stenosis. We determined velocity cutoffs for Ն70% stenosis on digital subtraction angiography by Warfarin-Aspirin Symptomatic Intracranial Disease criteria and evaluated novel stenotic/prestenotic ratio and low-velocity criteria. Results-A total of 102 patients with intracranial atherosclerotic disease (age 57Ϯ13 years; 72% men; median NationalInstitutes of Health Stroke Scale 3, interquartile range 6) provided 690 transcranial Doppler/digital subtraction angiography vessel pairs. On digital subtraction angiography, Ն50% stenosis was found in 97 and Ն70% stenosis in 62 arteries. Predictive values for transcranial Doppler SONIA criteria were similar (PϾ0.9) between middle cerebral artery (sensitivity 78%, specificity 93%, positive predictive value 73%, negative predictive value 94%, and overall accuracy 90%) and vertebral artery/basilar artery (69%, 98%, 88%, 93%, and 92%). As a single velocity criterion, most sensitive mean flow velocity thresholds for Ն70% stenosis were: middle cerebral artery Ͼ120 cm/s (71%) and vertebral artery/basilar artery Ͼ110 cm/s (55%). Optimal combined criteria for Ն70% stenosis were: middle cerebral artery Ͼ120 cm/s, or stenotic/prestenotic ratio Ն3, or low velocity (sensitivity 91%, specificity 80%, receiver operating characteristic 0.858), and vertebral artery/basilar artery Ͼ110 cm/s or stenotic/prestenotic ratio Ն3 (60%, 95%, 0.769, respectively). Conclusions-At laboratories with a standardized scanning protocol, SONIA mean flow velocity criteria remain reliably predictive of Ն50% stenosis. Novel velocity/ratio criteria for Ն70% stenosis increased sensitivity and showed good agreement with invasive angiography. (Stroke. 2011;42:3429-3434.)