1. Vasoconstrictor responses of the isolated and perfused canine epicardial coronary artery to uridine 5'-triphosphate (UTP) were analysed pharmacologically. 2. At basal perfusion pressure, UTP induced vasoconstriction in a dose-related manner and the vasoconstriction was sometimes followed by a slight vasodilatation at large doses (more than 10 nmol). The rank order of potency for vasoconstriction was UTP = UDP > ATP > TTP > or = ITP >> UMP. At raised perfusion pressure by 20 mM KCl, the vasoconstriction was not changed and a small vasodilatation was induced at large doses. The rank order of potency for vasodilatation was induced at large doses. The rank order of potency for vasodilatation was ATP >> ITP > or = UDP > UTP > or = TTP. The maximal vasodilator response to UTP was much less than that to ATP. UMP did not induce vasodilatation. 3. The P2X receptor agonist and desensitizing agent alpha, beta-methylene ATP (1 microM) and the P2 receptor antagonist suramin (100 microM) inhibited the vasoconstrictor responses to ATP but not those to UTP and UDP. The P2 receptor antagonist reactive blue 2 (30 microM) did not inhibit the vascular responses to UTP. 4. UTP (200 microM) desensitized the vasoconstrictor responses to UTP, but not either the vasodilator responses to UTP or the vasoconstrictor responses to ATP and UDP. UDP (200 microM) did not desensitize the vascular responses to UTP. 5. Preincubating the UDP stock solution and arterial preparation with hexokinase (10 and 1 uml-1, respectively) did not change the vasoconstrictor responses to UDP. 6. The Ca channel blocker diltiazem (1 microM) inhibited the vasoconstrictor responses to UTP but not those to ATP and UDP. Incubation in a Ca(2+)-free solution containing 1 mM EGTA inhibited the vascular responses to ATP, UTP and UDP. 7. Removal of the endothelium by an intraluminal injection of saponin (1 mg) inhibited the vasodilator responses to UTP. Indomethacin, a cyclo-oxygenase inhibitor (1 microM), inhibited the vasodilator responses to UTP, but NG-nitro-L-arginine, a nitric oxide synthase inhibitor (300 microM), did not have an inhibitory effect. 8. The results suggest that (1) UTP induces vasoconstriction via UTP-preferring P2Y receptors on the smooth muscle and vasodilatation via receptors different from those mediating the vasoconstriction induced by UTP and mediating the vasodilatation by ATP on the endothelium, through mainly the release of prostacyclin in the canine epicardial coronary artery; (2) UDP induces vasoconstriction via UDP-preferring P2Y receptors; and (3) L-type Ca ion channels are involved in the vasoconstriction induced by UTP, but not in that induced by UDP.