If adenosine (ADO) mediates metabolic vasodilation in the heart, increases in interstitial ADO (ISF[ADO]) must accompany increases in coronary vascular conductance. We tested this using ADO release, defined as the difference in [ADO] in coronary venous and arterial plasma multiplied by coronary plasma flow, as an index of ISF[ADO]. Pentobarbital-anesthetized dogs received intravenous norepinephrine or left atrial pacing, and the resulting changes in coronary blood flow (delta CBF), conductance (delta C), myocardial oxygen consumption (delta VO2), and ADO release (delta RADO) were measured. If ISF[ADO] and C are coupled, the ratio delta RADO/delta C should be greater than zero. For dogs receiving atrial pacing, the ratios delta RADO/delta C, delta RADO/delta CBF, and delta RADO/delta VO2 equal -2.4 +/- 2.2 nmol . mmHg-1 . ml-1, -0.022 +/- 0.020 nmol/ml, and -0.13 +/- 0.12 nmol/ml, respectively. These values do not differ from zero. For dogs receiving norepinephrine, delta RADO/delta C, delta RADO/delta CBF, and delta RADO/delta VO2 equal 9.7 +/- 1.8, 0.051 +/- 0.017, and 0.44 +/- 0.13, respectively. These values are greater than zero (P less than 0.05). These differences between atrial pacing and norepinephrine infusion rate observed despite similar changes in C, CBF, and VO2. We conclude that ADO may mediate the vasodilation induced by norepinephrine, but not atrial pacing.