We describe a technique to predict preoperatively the safety of permanently occluding an internal carotid artery. The method was performed by imaging stable xenon cerebral blood flow (CBF) with the internal carotid artery both open and temporarily occluded with a nondetachable balloon on a double lumen Swan-Ganz catheter. Patients were those in whom we planned to sacrifice the internal carotid artery (those with giant or inaccessible aneurysms) or those in whom such a sacrifice was at least likely (those with skull base tumors). Patients were divided into three groups on the basis of a comparison of occluded and nonoccluded CBF values. Group-I patients had no significant change in CBF with internal carotid artery occlusion; group-II patients showed a symmetric decrease in CBF; and group-III patients had an asymmetric decrease in CBF, always greater on the occluded side. A fourth group clinically failed to tolerate even brief carotid occlusion. The internal carotid artery in one patients from group III was sacrificed at surgery: the size and shape of his postoperative infarct corresponded almost exactly to the area of asymmetrically decreased CBF on his occluded study. The data suggest that if surgery is likely to result in permanent occlusion of the internal carotid artery, then patients who are at risk for delayed neurologic injury due to a compromised cerebral blood flow should have arterial bypass grafts before such surgery is performed.