We compared the rates of recanalization cerebral infarct and hemorrhage between intra-arterial (i.a.) reteplase and intravenous (i.v.) alteplase thrombolysis in a canine model of basilar artery thrombosis. Thrombosis was induced by injecting a clot in the basilar artery of 13 anesthetized dogs via superselective catheterization. The animals were randomized in a blinded fashion, 2 h after clot injection and verification of arterial occlusion, to receive i.v. alteplase 0.9 mg/kg over 60 min and i.a. placebo, or i.a. reteplase 0.09 units/kg over 20 min, equivalent to one-half the alteplase dose, and i.v. placebo. Recanalization was studied for 6 h after treatment with serial angiography; the images were later graded in a blinded fashion. Blinded interpretation of postmortem MRI was performed to assess the presence of brain infarcts and/or hemorrhage. At 3 h after initiation of treatment, partial or complete recanalization was observed in one of six dogs in the i.v. alteplase group and in five of seven in the i.a. reteplase group (P = 0.08). At 6 h, no significant difference in partial or complete recanalization was observed between the groups (two of six vs. five of seven; P = 0.20). Postmortem MRI revealed infarcts in four of six animals treated with i.v. alteplase and three of seven treated with i.a. reteplase (P = 0.4). Intracerebral hemorrhage was more common in the i.v. alteplase group (four of six vs. none of seven; P = 0.02). This study thus suggests that i.a. thrombolysis affords a recanalization rate similar to that of i.v. thrombolysis, but with a lower rate of intracerebral hemorrhage.