Conclusion: Color duplex ultrasound has high sensitivity and specificity for diagnosis of spontaneous internal carotid artery dissection that causes neurologic symptoms. Summary: Carotid duplex ultrasound is frequently used in evaluation of patients with possible carotid artery dissection. The authors sought to determine the overall accuracy of color duplex ultrasound to diagnose spontaneous internal carotid artery dissection in patients with a first episode of carotid distribution ischemia. Patients with a first episode of carotid territory stroke, transient ischemic attack, or retinal ischemia underwent an electrocardiogram (ECG), color duplex ultrasound of the cervical carotid arteries, and clinical and hematologic examinations. Brain computed tomography (CT) scanning was performed in individuals with stroke or transient ischemic attacks, and echocardiography and 24-hour ECG monitoring were used in selective cases. Patients were included if they were <65 years, color duplex ultrasound showed a probable internal carotid artery dissection (stenosed or occluded internal carotid artery), or they had no determined etiology of cerebral ischemia. Included patients also underwent cervical magnetic resonance imaging (MRI) and MR angiography (MRA) and possibly cerebral catheter angiography as well. Those who performed the color duplex ultrasound studies were blinded to the results of the MRI and angiographic studies. The study screened 1652 patients, and 177 were included. Of the 1475 excluded patients, 818 were >65 years, 1081 had another cause of carotid artery ischemia, and 58 had intercranial hemorrhage. Internal carotid artery dissection was diagnosed by carotid duplex ultrasound in 77 of the 177 patients. The etiology of ischemia was undetermined in the remaining 100 patients. Cervical MRI and MRA indicated 74 of the 177 patients had spontaneous internal carotid artery dissection. There were three false-negative and six false-positive color duplex ultrasound examinations. Therefore, sensitivity, specificity, positive, and negative predictive values for diagnosis of spontaneous internal carotid artery dissection using carotid artery duplex ultrasound in patients with carotid territory ischemic symptoms were 96%, 94%, 92%, and 97%, respectively. Comment: The very high negative predictive value for excluding spontaneous carotid artery dissection with color duplex ultrasound essentially allows withholding of treatment for internal carotid artery dissection in patients in which that diagnosis is suspected but in whom the duplex ultrasound shows no evidence of dissection. Positive predictive value is also excellent, but because such patients will be treated with anticoagulation, or perhaps other interventions, patients in whom internal carotid artery dissection is diagnosed by duplex should have a confirmatory study before treatment is initiated.