Little information exists on the utility of transcranial Doppler sonography (TCD) in detecting anterior (ACA) and posterior cerebral artery (PCA) vasospasm following subarachnoid hemorrhage. During the period at risk for vasospasm, 53 patients with subarachnoid hemorrhage who had technically adequate TCD performed within 24 hours of cerebral angiography, allowing evaluation of 87 ACAs and 84 PCAs, were studied. ACA and PCA vasospasm were defined by mean blood flow velocities of at least 120 cm/sec and at least 90 cm/sec, respectively. For detection of ACA vasospasm, sensitivity was 18% and specificity was 65%. For PCA vasospasm, sensitivity was 48% and specificity was 69%. False-positive findings for occlusion accounted for 12 (92%) of 13 ACA of false-positive results and 7 (37%) of 19 PCA false-positive results, and were most often due to anatomical factors and operator error or inexperience. After exclusion of both true-positive and false-positive findings for occlusion and changes in the diagnostic criterion to at least 130 cm/sec for ACA vasospasm and at least 110 cm/sec for PCA vasospasm, specificity improved for both types of vasospasms (100 and 93%, respectively). However, the sensitivity of TCD to detect ACA and PCA vasospasm is limited by a variety of anatomical, technical, and other factors. It is concluded that TCD is highly specific in detecting both ACA and PCA vasospasm on arteries that can be insonated.