1. Isolated hearts of guinea-pigs were perfused in vitro with a physiological salt solution via a retrograde aortic cannulation (Langendorff preparation) at constant perfusion pressure. Bolus intra-arterial injections of various vasodilator drugs were made and the coronary flow responses were measured with an electromagnetic flow probe placed in the arterial inflow circuit. Inhibitory drugs were infused intra-arterially. 2. Nitro-L-arginine (NLA; 500 microM), an NO synthesis inhibitor, decreased coronary baseline flow by 16 +/- 0.8%, converted acetylcholine-induced coronary vasodilatation to vasoconstriction and had no effect on coronary flow responses to adenosine or papaverine. Sodium nitroprusside-induced responses were enhanced during NLA infusion by 46 +/- 11%. 3. Adenosine 5'-triphosphate (ATP) increased coronary flow but coronary flow responses to ATP were not altered by infusion of NLA. 4. ATP-induced coronary dilatation was not significantly attenuated by infusion of the adenosine receptor antagonist XAC, (xanthine amine congener; 2 microM), whereas XAC decreased coronary flow responses to adenosine by 75% +/- 5%. 5. ATP-induced coronary flow responses were reduced by only 31 +/- 4% during indomethacin infusion (2.8 microM) whereas indomethacin completely eliminated the initial vasoconstriction phase and greatly attenuated the peak flow and duration of the later vasodilatation phase seen in response to arachidonic acid (0.75 nmol). Indomethacin had no effect on vasodilatations produced by adenosine or prostaglandin I2. 6. These results indicate that ATP-induced coronary dilatation in the isolated, perfused heart of the guinea-pig is not dependent upon NO production or upon degradation of ATP to adenosine. The coronary dilator action of ATP may be partially dependent (approximately 30%) upon the production of vasodilator prostaglandins.