Surgical sacrifice of the carotid artery is a frequently anticipated event during the treatment of certain aneurysms and tumors. The ability to predict tolerance to carotid artery occlusion is therefore of benefit when planning procedures in which the carotid artery is at risk. A trial of carotid artery occlusion using an angiographic balloon during concurrent neurological examination or blood flow studies is an accepted method for testing tolerance, but it carries the risks of an angiogram and cannot be performed at the bedside. Transcranial Doppler ultrasound (TCD) is a noninvasive modality that permits measurement of blood velocity in cerebral vessels. The immediate effects of carotid artery occlusion on middle cerebral artery (MCA) perfusion can be obtained by insonating this artery during manual carotid artery compression. To compare the TCD response to carotid artery compression with the data obtained with more formal testing, the MCA of 22 patients was insonated during manual carotid artery compression and the results compared with the clinical tolerance to balloon occlusion in all patients and to blood flow studied by single photon emission computerized tomography before or during balloon occlusion in 14 of the 22 patients. Surgery was planned to treat giant unruptured aneurysms in 17 cases, intracranial tumors in three, a carotid-cavernous fistula in one, and a carotid artery injury in one. Fifteen patients showed a reduction in TCD flow velocities by no more than 65%; of these, 14 (93%) clinically tolerated the balloon occlusion test. Of the seven patients showing a TCD flow velocity decrease of more than 65%, six (86%) developed a transient focal deficit during the occlusion. It is concluded that the change in MCA velocity measured with TCD studies during manual carotid artery occlusion is a useful predictor of the clinical and blood flow responses to a trial of carotid artery occlusion with an angiographic balloon.