Although recent studies suggest that erythropoietin (EPO) may reduce multiple features of the myocardial ischemia/reperfusion injury, the cellular mechanisms and the clinical implications of EPO-induced cardioprotection are still unclear. Thus, in this study, we clarified dose-dependent effects of EPO administered just before reperfusion on infarct size and the incidence of ventricular fibrillation and evaluated the involvement of the phosphatidylinositol-3 (PI3) kinase in the in vivo canine model. The canine left anterior descending coronary artery was occluded for 90 min followed by 6 h of reperfusion. A single intravenous administration of EPO just before reperfusion significantly reduced infarct size (high dose (1,000 IU/kg): 7.7 +/- 1.6%, low dose (100 IU/kg): 22.1 +/- 2.4%, control: 40.0 +/- 3.6%) in a dose-dependent manner. Furthermore, the high, but not low, dose of EPO administered as a single injection significantly reduced the incidence of ventricular fibrillation during reperfusion (high dose: 0%, low dose: 40.0%, control: 50.0%). An intracoronary administration of a PI3 kinase inhibitor, wortmannin, blunted the infarct size-limiting and anti-arrhythmic effects of EPO. Low and high doses of EPO equally induced Akt phosphorylation and decreased the equivalent number of TUNEL-positive cells in the ischemic myocardium of dogs. These effects of EPO were abolished by the treatment with wortmannin. In conclusion, EPO administered just before reperfusion reduced infarct size and the incidence of ventricular fibrillation via the PI3 kinase-dependent pathway in canine hearts. EPO administration can be a realistic strategy for the treatment of acute myocardial infarction.