Postischemic myocardial contractile dysfunction is in part mediated by the burst of reactive oxygen species (ROS), which occurs with the reintroduction of oxygen. We hypothesized that tissue oxygen tension modulates this ROS burst at reperfusion. After 20 min of global ischemia, isolated rat hearts were reperfused with temperature-controlled (37.4 degrees C) Krebs-Henseleit buffer saturated with one of three different O2 concentrations (95, 20, or 2%) for the first 5 min of reperfusion and then changed to 95% O2. Additional hearts were loaded with 1) allopurinol (1 mM), a xanthine oxidase inhibitor, 2) diphenyleneiodonium (DPI; 1 microM), an NAD(P)H oxidase inhibitor, or 3) Tiron (10 mM), a superoxide scavenger, and were then reperfused with either 95 or 2% O2 for the first 5 min. ROS production and tissue oxygen tension were quantitated using electron paramagnetic resonance spectroscopy. Tissue oxygen tension was significantly higher in the 95% O2 group. However, the largest radical burst occurred in the 2% O2 reperfusion group (P < 0.001). Recovery of left ventricular (LV) contractile function and aconitase activity during reperfusion were inversely related to the burst of radical production and were significantly higher in hearts initially reperfused with 95% O2 (P < 0.001). Allopurinol, DPI, and Tiron reduced the burst of radical formation in the 2% O2 reperfusion groups (P < 0.05). Hypoxic reperfusion generates an increased ROS burst originating from multiple pathways. Recovery of LV function during reperfusion is inversely related to this oxygen radical burst, highlighting the importance of myocardial oxygen tension during initial reperfusion.