In preoperative embolization of cerebral arteriovenous malformation (AVM), one difficulty frequently encountered is assessment of the risk of ischemic deficit, posed by proceeding with embolization. One technique to assist prediction of tolerance of devascularization involves superselective injection of sodium amytal into cerebral arteries, prior to and during embolization. Since September, 1985, we have performed 119 embolization procedures in 77 patients with cerebral AVM. Of these, 89 procedures in 60 patients involved superselective intracerebral catheterization and embolization with either polyvinyl alcohol (PVA) alone or in combination with platinum microcoils, surgical silk, or a mixture of ethanol and microfibrillar collagen. In 50 of these procedures, we superselectively administered sodium amytal intraarterially immediately prior to embolization, for purposes of functional testing. Groups of patients undergoing testing were compared to groups not tested prior to embolization, for development of transient or permanent neurologic deficits or cerebral hemorrhage, subsequent to embolization. Significantly better results, both in absolute numbers and severity of complications, were found when testing was employed. No complications of amytal administration were seen. In four cases, embolization was altered or discontinued based on development of a deficit at the time of amytal administration. We conclude that the brief anesthetic effect of intraarterial testing with sodium amytal is a safe and important adjunct during preoperative embolization of cerebral AVMs.