It is generally accepted that occlusion of a major coronary artery in the dog results in a transmural gradient of collateral blood flow, with the subepicardial region receiving the greatest perfusion. The lateral and base to apex distribution of collateral blood flow and of metabolic and functional consequences of ischemia have been more difficult to define. One reason for such difficulties has been the failure to define the anatomic boundaries of the ischemic vascular bed so that uncontaminated samples of ischemic and non-ischemic tissue could be selected for study. In the present study, the three dimensional distribution of myocardial blood flow during occlusion of the circumflex artery was mapped in seven dogs. At the end of the study, the boundaries of previously ischemic and non-ischemic regions were identified by simultaneous coronary perfusion with red and blue dyes. Left ventricular slices were separated into ischemic and non-ischemic vascular beds based on the dye boundaries, with 1-2 mm of tissue trimmed from this interface to eliminate visually apparent admixture. The ischemic vascular bed of each cross sectional slice then was cut into five transmural wedges, each 3-5 mm wide; each wedge was further subdivided into subendocardial, middle, and subepicardial thirds. The results of blood flow measurements in these samples indicate that the dye injection technique identifies a real interface with a sharp lateral transition in blood flow between ischemic and non-ischemic vascular beds. Within the ischemic vascular bed, there is a transmural gradient of collateral blood flow, but within a given mural layer, there is no consistent gradient from the center to lateral edge or from base to apex of the ischemic region. Thus, in studies designed to characterize the properties of myocardium on either side of the ischemic/non-ischemic interface, reasonable resolution can be achieved by coronary dye infusions to permit visual identification of this interface. On the other hand, in studies in which collateral blood flow is measured as a baseline predictor of infarct size, measurements can be made in a central ischemic block which will be representative of most or all of the ischemic region. Borderzone samples can be excluded to avoid contamination of ischemic samples with non-ischemic tissue.