Contractile dysfunction of viable, previously ischemic stunned myocardium is thought to be due to reactive oxygen species generated during ischemia/reperfusion. Direct in vivo evidence that oxidants cause systolic or diastolic dysfunction of viable myocardium has, however, been lacking. We sought to determine whether in vivo exposure of canine myocardium to exogenously generated reactive oxygen species could--in the absence of myocardial ischemia or necrosis--"mimic" the depressed systolic contractile function, paradoxical contraction during early diastole, and prolonged diastolic relaxation time characteristic of stunned myocardium. Anesthetized open-chest dogs were randomly assigned to receive either (1) the free radical generating substrates xanthine oxidase + purine + iron-saturated transferrin or (2) saline, infused directly into an anterior coronary vein. Infusion of free radical substrates did not cause ischemia: regional myocardial blood flow and myocardial high-energy phosphate stores were normal in both groups. Furthermore, infusion of xanthine oxidase + purine + transferrin was not associated with histologic or electron microscopic evidence of myocyte injury or death in this model. Xanthine oxidase + purine + transferrin did, however, produce marked abnormalities in regional systolic contractile function; at 2 hours after the onset of infusion, segment shortening (assessed by sonomicrometry) in the perfused region of the heart averaged 62 +/- 5% of baseline, preinfusion values in animals infused with free radical substrates versus 113 +/- 8% of baseline values in saline-administered control dogs (p less than 0.004). This systolic dysfunction was effectively reversed by administration of the free radical scavenging agents superoxide dismutase + catalase.(ABSTRACT TRUNCATED AT 250 WORDS)