The role of microvascular damage in the genesis of the “no-reflow” phenomenon was investigated in the left ventricular myocardium of dogs subjected to temporary occlusions of a major coronary artery for 40 and 90 min. Intravenous carbon black or thioflavin S (a fluorescent vital stain for endothelium) were used to demonstrate the distribution of coronary arterial flow in control and damaged myocardium. These tracers were injected simultaneously with release of the coronary occlusion or after 5 or 20 min of reflow of coronary arterial blood. After 40 min of ischemia plus arterial reperfusion, usually the tracers were evenly distributed throughout the damaged tissue at each time of reperfusion. On the other hand, when reflow was allowed after 90 min of ischemia, portions of the inner half of damaged myocardium were not penetrated by the tracers. Electron microscopic study of this poorly perfused tissue revealed severe capillary damage; endothelial cells with large intraluminal protrusions and decreased pinocytic vesicles were common. Also, occasional intraluminal fibrin thrombi were noted, as well as extravascular fibrin deposits and erythrocytes. Myocardial cells were swollen in both poorly perfused and well-perfused irreversibly injured tissue. Contraction bands and mitochondrial Ca²⁺ accumulation were prominent features of irreversible injury with reflow at 40 min but were not noted after 90 min of ischemia in areas with poor perfusion. These results suggest that 40 min of ischemia were tolerated by the capillary bed of the dog heart without serious capillary damage or perfusion defects, but that 90 min of ischemic injury was associated with the “no-reflow” phenomenon, i.e., failure to achieve uniform reperfusion. This failure of reflow was associated with extensive capillary damage and myocardial cell swelling. Death of severely ischemic myocardial cells in this model occurs before the onset of capillary damage and the no-reflow phenomenon.