The mechanism of coronary vasodilation produced by exercise is not understood completely. Recently, we reported that blockade of vascular smooth muscle KAT channels decreased coronary blood flow at rest, but did not attenuate the increments in coronary flow produced by exercise. Adenosine is not mandatory for maintaining basal coronary flow, or the increase in flow produced by exercise during normal arterial inflow, but does contribute to coronary vasodilation in hypoperfused myocardium. Therefore, we investigated whether adenosine opposed the hypoperfusion produced by KA+P channel blockade, thereby contributing to coronary vasodilation during exercise. 11 dogs were studied at rest and during exercise under control conditions, during intracoronary infusion of the KA channel blocker glibenclamide (50 ,ug/kg per min), and during intracoronary glibenclamide in the presence of adenosine receptor blockade. Glibenclamide decreased resting coronary blood flow from 45±5 to 35±4 ml/min (P < 0.05), but did not prevent exerciseinduced increases of coronary flow. Glibenclamide caused an increase in myocardial oxygen extraction at the highest level of exercise with a decrease in coronary venous oxygen tension from 15.5±0.7 to 13.6±0.8 mmHg (P < 0.05). The addition of the adenosine receptor antagonist 8-phenyltheophylline (5 mg/kg intravenous) to KAT channel blockade did not further decrease resting coronary blood flow but did attenuate the increase in coronary flow produced by exercise. This was accompanied by a further decrease of coronary venous oxygen tension to 10.1±0.7 mmHg (P < 0.05), indicating aggravation of the mismatch between oxygen demand and supply. These findings are compatible with the hypothesis that KATP channels modulate coronary vasomotor tone both under resting conditions and during exercise. However, when KA channels are blocked, adenosine released from the hypoperfused myocardium provides an alternate mechanism to mediate coronary vasodilation in response to increases in oxygen demand produced by exercise. (J. Clin. Invest. 1995. 95:285-295.)