To examine the hemodynamic interdependence of coronary arteries and veins, great cardiac vein (GCV) and left anterior descending (LAD) coronary artery flows of dogs were measured. Although they were not different under base-line conditions, 42.9 +/- 4.1 and 34.9 +/- 5.2 ml/min, respectively, LAD occlusion caused only a 56% decrease in GCV flow, whereas peak GCV flow during reactive hyperemia consistently underestimated peak LAD flow by nearly 40%. The measured hemodynamic data were very closely predicted by a simple model that assumed the presence of a low-resistance venous collateral anastomotic pathway between the GCV and veins outside the LAD territory. The resistance of the venous collaterals averaged 9.9 +/- 3.0% of LAD resistance. Therefore, GCV flow does not solely represent LAD collateral flow when the LAD is occluded, nor is it a reliable indicator of changes in LAD perfusion patterns. Nutritional myocardial perfusion assessed with 133Xe washout decreased from 0.86 +/- 0.16 to 0.13 +/- 0.02 ml.min-1.g-1 after LAD occlusion, whereas retroperfusion of the GCV with arterial blood enhanced washout by 85% to 0.23 +/- 0.03 ml.min-1.g-1 (P less than 0.005). Retroperfusion was inefficient, however, since only 10% of the isotope injected into the GCV gained access to the myocardium.