1 Adenosine-5'-triphosphate (ATP) and adenosine are potent coronary vasodilators. ATP is rapidly converted to adenosine by ectonucleotidases. We examined whether coronary vasodilation caused by exogenous ATP is mediated by P 2 receptor activation or by A 2A -adenosine receptor activation. 2 E ects of interventions on coronary conductance were determined by measuring coronary perfusion pressure in guinea-pig isolated hearts perfused at a constant¯ow of 10 ml min 71 . 3 ATP and adenosine both caused sustained, concentration-dependent increases of coronary conductance. Maximal responses to both agonists were equivalent. The values of pD 2 (+s.e.mean) for ATP and adenosine were 6.68+0.04 and 7.06+0.05, respectively. Adenosine was signi®cantly more potent than ATP (P50.0001, n=10).4 The values of pIC 50 for the selective A 2A -adenosine receptor antagonist SCH58261 to antagonize equivalent responses to ATP and adenosine were 8.28+0.08 and 8.28+0.06 (P=0.99, n=6), respectively. 5 The non-selective adenosine receptor antagonists xanthine amine congener (XAC) and CGS15943 antagonized similarly the equivalent vasodilations caused by ATP (pIC 50 values 7.48+0.04 and 7.45+0.06, respectively) and adenosine (pIC 50 values 7.37+0.13 and 7.56+0.11). 6 In contrast to ATP and adenosine, the two P 2 agonists 2-methylthio-ATP and uridine-5'-triphosphate failed to cause stable increases of coronary conductance, caused desensitization of vasodilator responses, and were not antagonized by SCH 58261, 8-parasulphophenyltheophylline, or XAC. 7 Glibenclamide attenuated coronary vasodilations caused by ATP and adenosine by 88 and 89%, respectively, but failed to attenuate those caused by 2-methylthio-ATP.