The "no-reflow" phenomenon, the occurrence of areas with very low flow in hearts reperfused after ischemia, is thought to be largely established at the time of reperfusion as a result of microvascular damage induced by ischemia. In the present study we sought to determine whether additional impairment of tissue perfusion might also occur during the course of reperfusion. Open-chest dogs were subjected to 90 minutes of left circumflex coronary artery occlusion and reperfused for 2 minutes (n=7) or 3.5 hours (n=8). Myocardial perfusion was visualized in left ventricular slices following in vivo injection of the fluorescent dye thioflavin-S just before killing. The area of impaired perfusion (absent thioflavin) averaged 9.5+3.0% of the risk region in dogs reperfused for 2 minutes, whereas it was nearly three times as large in dogs reperfused for 3.5 hours (25.9 ±8.2% of the risk region, p<0.05). Serial measurements of flow by microspheres during reperfusion demonstrated zones within the postischemic myocardium that were hyperemic 2 minutes after reperfusion, with adequate flow still present at 30 minutes, but with a subsequent marked fall in perfusion. After 3.5 hours these areas showed negligible flow (0.13 ±0.03 ml/min/g) and no thioflavin uptake. Tissue samples showving postischemic impairment in perfusion had received virtually no collateral flow during ischemia (<0.01 mlUmin/g), whereas collateral flow was significantly higher in adjacent thioflavin-positive zones (0.04±0.01 ml/min/g in endocardial samples and 0.07±0.02 ml/min/g in samples from the midmyocardium, p<0.001 vs. thioflavin-negative areas). Areas that showed late impairment of flow invariably demonstrated contraction band necrosis, which contrasted with the pattern of coagulation necrosis observed in areas of "true" (i.e., immediate) no-reflow. Intracapillary erythrocyte stasis and marked intravascular neutrophil accumulation (to levels >20-fold that found after 2 minutes reperfusion) were typically observed in areas of delayed impairment to flow. Obstruction to flow at the capillary level was confirmed in additional dogs in which the heart was injected postmortem with silicone rubber to delineate the microvascular filling pattern. Areas of absent capillary filling were much more extensive after 3.5 hours than after 2 minutes reperfusion. Thus, this study shows that the occurrence of areas of markedly impaired perfusion in postischemic myocardium is related only in part to an inability to reperfuse certain areas on reflow. A more important factor is represented by a delayed, progressive fall in flow to areas that initially received adequate reperfusion. This phenomenon develops in regions receiving no collateral flow during ischemia and is associated with neutrophil accumulation and capillary plugging late during the course of reperfusion. (Circulation