A canine model was developed to investigate coronary artery thrombolysis and reocclusion in the setting of endothelial cell damage and fixed stenosis, which simulate anatomic features occurring in patients with acute myocardial infarction. In open chest dogs, endothelial cell damage was produced in the left anterior descending coronary artery by external compression with blunt forceps, >90% stenosis was obtained by an external constrictor and thrombosis was induced by instillation of thrombin and fresh blood in an isolated arterial segment. In the absence of stenosis, intravenous infusion of 750,000 U of streptokinase over 1 h caused reperfusion in five of six dogs in 34 ± 25 min (mean ± SD). Urokinase, 600,000 U intravenously over 30 min followed by 600,000 U over 30 min by the intracoronary route, induced reperfusion in three of four dogs in 65 ± 23 min. Recombinant two chain tissue-type plasminogen activator (rt-PA) (G11021), infused intravenously at a rate of 15 μg/kg per min for 30 min or until reflow, induced reperfusion in all 12 dogs in 28 ± 13 min. In the absence of coronary artery stenosis, spontaneous reocclusion did not occur within 2 h after the end of the infusion. In the presence of the coronary artery constrictor, which reduced the blood flow to 40 ± 10% of baseline, streptokinase, urokinase and rt-PA caused coronary thrombolysis to proceed at comparable or only slightly slower rates. Cyclical reocclusion during or after the end of infusion of these thrombolytic agents, caused by platelet-rich thrombus, was almost invariably observed, generally within 30 min after the onset of reperfusion. It is concluded that this canine model simulates several characteristic features of coronary artery thrombosis in human patients and, therefore, appears to be useful for the study of both the potency of thrombolytic agents and strategies for the prevention of reocclusion after coronary artery thrombolysis.