BACKGROUND: The prevalence of subclavian steal (defined as retrograde/bidirectional vertebral artery flow) in the general population and in patients undergoing cerebrovascular duplex ultrasound (CDUS) examinations is variable. This is the largest study to date to analyze the incidence of duplex-suggested subclavian steal in 5615 CDUS examinations over a 1-year period and to examine its clinical implications. PATIENT POPULATION AND METHODS: All consecutive CDUS examinations performed over a 1-year period were analyzed for the presence of subclavian steal. Indications of testing, presence of posterior cerebral circulation/subclavian steal symptoms, and any interventions for subclavian steal were analyzed. RESULTS: A total of 171 of 5615 (3.1%) CDUS examinations were found to have subclavian steal (duplex-suggested). One hundred seventeen (2.1%) had retrograde flow and 54 (1%) had bidirectional flow. Of 171, 104 (60.8%) were left sided. Indications for CDUS were post-carotid endarterectomy/carotid artery stenting surveillance in 39 patients (22.8%), surveillance for progression of carotid stenosis in 76 patients (44.4%), transient ischemic attack/stroke in 26 patients (15%), asymptomatic screening/carotid bruit in 18 patients (10.5%), and isolated posterior cerebral circulation symptoms in 12 patients (7%). A total of 63% patients had associated >50% carotid stenosis. The mean arm Doppler pressure gradient was 32.2 mm Hg for asymptomatic patients vs 37 mm Hg for patients with posterior circulation symptoms (P = .3254). There were significant differences between the mean systolic arm pressure for patients with retrograde vs antegrade vs bidirectional flow (105 mm Hg vs 146 mm Hg vs 134 mm Hg, respectively, P < .0001). All patients with retrograde flow had >50% subclavian stenosis or occlusion (100 of 117 had subtotal/total occlusion) except for one patient. Meanwhile, 52 of 54 patients with bidirectional flow had >50% subclavian stenosis (6 of 54 with subtotal/total occlusion), whereas two patients were normal/<50% stenosis (P < .0001). Overall, 26 of 171 patients (15.2%) had interventions for disabling symptoms. Eleven of 26 of all interventions were for disabling arm claudication, and only 10 of 171 patients (5.8%) were intervened for disabling posterior circulation symptoms with complete resolution of symptoms in all except one. At a late follow-up with a mean of 18 months (range: 1-37 months), there was no late major stroke with only two lacunar infarcts (not subclavian steal related). There were also seven late deaths, none stroke related. CONCLUSIONS: The incidence of subclavian steal in patients who undergo CDUS is relatively rare. Most of these patients are asymptomatic and can be treated conservatively, and only a few may need intervention for disabling symptoms with good symptom resolution.