We measured epicardial S-T segment elevation and blood flow to the subjacent myocar-dium in 8-12 samples from anesthetized, open-chest dogs with coronary occlusion. The response to iv adenosine of blood flow to “ischemic” myocardium/mean aortic pressure depended on which criterion (among criteria I-V) was used to select samples as “ischemic”: I—(S-T ⩾ 2 mV), 125 ± 47% increase(P< 0.08); II—(S-T ⩾ 4 mV), 58 ± 48% increase [not significant (NS)]; III—(myocardial blood flow < 50% of control zone), 6 ± 19% decrease (NS); IV—(flow < 25% of control), 13 ± 16% decrease (NS); V—(< 2% normal zone tissue, labeled by a special microsphere technique), 46 ± 8% decrease(P< 0.03). Peripheral coronary pressure and retrograde flow/mean aortic pressure decreased by 23 and 28%, respectively(P< 0.05), to suggest that criterion V was correct in identifying as ischemic a group of samples in which adenosine decreased collateral flow. The higher apparent control collateral flow values (criteria I-III) and unchanged or increased flow to samples of supposedly ischemic myocardium (criteria I-IV) could be explained by their larger fractions of normal zone tissue mixed in with ischemic tissue: absolute flow change during adenosine = 3.85 (fraction of normal zone tissue in sample)—0.156 ml/min per g,r= 0.94,n =64. We conclude that criteria which fail to exclude contamination of supposedly ischemic samples by normal zone tissue can lead to misinterpretation of the direction and/or magnitude of changes in collateral flow to ischemic myocardium during vasodilation. Changes in flow to samples containing mixtures of normal and ischemic tissue may not be relevant to attempts to salvage truly ischemic myocardium by increasing actual collateral myocardial blood flow.Circ Res 47:108-116,1980