The change in left ventricular ejection fraction from preintervention to predischarge was prospectively assessed in 393 patients with acute myocardial infarction. Within 12 h of symptom onset (mean 6.3 +/- 2.7 h), patients were randomly assigned to a double-blind intracoronary infusion of streptokinase, nitroglycerin, both streptokinase and nitroglycerin or conventional therapy without acute cardiac catheterization. Treatment effects were also assessed in prospectively defined angiographic subsets. There was a significant interaction between streptokinase and nitroglycerin (p less than 0.01), resulting in an increase in ejection fraction of 3.9 percentage units in the combined treatment arm (p less than 0.001). Patients with collateral flow to a totally obstructed infarct-related artery showed a significant improvement over those without collateral flow in the streptokinase (5.4 +/- 2.5%) and streptokinase-nitroglycerin (10.6 +/- 2.7%) arms, but not in the nitroglycerin arm. Time to treatment did not influence the change in ejection fraction. In patients with initial subtotal occlusion, thrombolytic therapy was of no short-term benefit because ejection fraction increased by 6% in all three intervention arms. These findings indicate that relatively late thrombolytic therapy results in significant myocardial salvage in those patients with collateralized total coronary occlusion. This benefit is potentiated by concomitant nitroglycerin therapy.