Endothelial damage or removal abolishes the dilation of epicardial coronary arteries induced by increments in flow through these arteries in vitro. Therefore, we tested whether or not the release of a cyclooxygenase product from endothelial cells in vivo is the mechanism of this flow-dependent dilation. In eight conscious dogs, instrumented to register the external diameter of two epicardial branches--anterior descending and circumflex--of the left coronary artery, increments in coronary flow increased and reductions in coronary flow decreased the diameter of the left circumflex epicardial artery by 182 +/- 11 micron/100% change in flow. When mean coronary flow in one epicardial branch was kept constant by a distal, flow-limiting stenosis during the application of flow-augmenting stimuli (temporal coronary occlusion or 80-400 micrograms/kg adenosine i.v.), no dilation of this artery was observed. Cyclooxygenase inhibition (suppressing the bradykinin-induced elevation of plasma 6-keto-PGF1 alpha) by indomethacin (5 mg/kg) or by diclofenac (10 mg/kg) increased smooth muscle tone in both epicardial arteries, but did not modify the flow-diameter relation (181 +/- 10 and 179 +/- 9 microns/100% change in flow, respectively). It is concluded that a tonic, instantaneous influence of coronary flow on the smooth muscle tone of the epicardial coronary arteries exists in vivo. It is unlikely that prostacyclin or another prostanoid is a mediator of this endothelium-mediated influence of flow on smooth muscle tone.