To try to unravel the complexity and heterogeneity of the "no-reflow" phenomenon and its underlying mechanisms, we studied tissue perfusion in reperfused heart muscle by using tracer microspheres in an anesthetized dog model of 90-minute coronary occlusion followed by reperfusion for 2 1/2 hours, 24 hours, or 1 week. Regional myocardial blood flow was determined both in basal flow conditions and during reactive hyperemia. The effect of intracoronary adenosine administration was examined, and the ultrastructure of postischemic myocardium was analyzed. In viable reperfused tissue (as delineated by triphenyltetrazolium chloride staining), reflow in basal conditions is unimpaired. Coronary flow reserve (as approximated by peak reactive hyperemic flow) is intact at the start of reperfusion, decreases by more than half after 2 1/2 hours, and recovers completely within 1 week. This impairment of coronary reserve can be relieved by intracoronary adenosine administration. On ultrastructural examination, the capillaries are patent. On the other hand, in irreversibly damaged myocardium, both the basal reflow impairment and the decrease in coronary flow reserve are severe and permanent. Coronary flow reserve is already decreased at the start of reperfusion, and the pharmacological intervention has no beneficial effect. Ultrastructurally, extracellular and intracellular edema invariably are present, whereas the vascular endothelium is damaged and the capillaries are packed with red blood cells. We conclude that the no-reflow phenomenon (i.e., mechanical obstruction to blood flow) is limited to infarcted tissue. In viable myocardium, however, coronary flow reserve is transiently diminished, probably because of washout and subsequent insufficient availability of the chemical mediator adenosine after breakdown and slow recovery of the precursor ATP pool.