To clarify the efficacy and limitations of the intra-arterial local infusion of a high-dose fibrinolytic agent for acute embolic stroke, we analyzed the results of 44 patients and compared them with those of 51 patients treated with intracarotid (18 patients) or intravenous (33 patients) infusion therapy. Ten megaunits of recombinant tissue plasminogen activator or 24 x 10(4) IU of urokinase were administered through a microcatheter placed into or proximal to an embolus for 20 minutes. When arterial recanalization was not achieved, a second or third infusion was performed. The rates of complete and partial recanalization just after the local infusion were 52 and 32%, respectively. They were high in middle cerebral and basilar artery occlusion and low in internal carotid artery occlusion (69, 78, and 20%, respectively). In our use, there was no difference between tissue plasminogen activator and urokinase in restoring blood flow. The mean time interval from onset to recanalization in patients with middle cerebral artery occlusion showing marked improvement was 4.8 hours, and it was 5.8 hours with basilar artery occlusion. The size of infarction was reduced, and the outcome was good in patients with complete recanalization achieved. The incidence of hemorrhagic infarction within 24 hours was 22%, and only one patient clinically deteriorated. In the intracarotid infusion group (20 x 10(4) IU of urokinase for 30 min), only two patients showed partial recanalization without clinical improvement. The incidence of hemorrhagic infarction was 28%. The outcome in this group and the intravenous infusion group (18 x 10(4) IU of urokinase a day for 1 wk) was poor compared with that in the local infusion group showing complete recanalization. This preliminary study appears to suggest that intra-arterial local fibrinolytic therapy could be a new strategy for acute embolic stroke.