1 The effects of the endocannabinoid, anandamide, and its metabolically stable analogue, methanandamide, on induced tone were examined in sheep coronary artery rings in vitro. 2 In endothelium-intact rings precontracted to the thromboxane A(2) Mimetic, U46619, anandamide (0.01-30 muM) induced slowly developing concentration-dependent relaxations (pEC(50) [negative log of EC50]=6.1 +/- 0.1; R-max [maximum response]=81 +/- 4%). Endothelium denudation caused a 10 fold rightward shift of the anandamide concentration-relaxation curve without modifying R-max. Methanandamide was without effect on U46619-induced tone. 3 The anandamide-induced relaxation was unaffected by the cannabinoid receptor antagonist, SR 141716A (3 muM), the vanilloid receptor antagonist, capsazepine (3 and 10 muM) or the nitric oxide synthase inhibitor, L-NAME (100 pm). 4 The cyclo-oxygenase inhibitor, indomethacin (3 and 10 muM) and the anandamide amidohydrolase inhibitor, PMSF (70 and 200 um), markedly attenuated the anandamide response. The anandamide transport inhibitor, AM 404 (10 and 30 pm), shifted the anandamide concentration-response curve to the right. 5 Precontraction of endothelium-intact rings with 25 mm KCI attenuated the anandamide-induced relaxations (R-max = 7 +/- 7%), as did K+ channel blockade with tetraethylammonium (TEA; 3 muM) or iberiotoxin (100 nm). Blockade of small conductance, Ca2+-activated K+ channels, delayed rectifier K+ channels, K-ATP channels or inward rectifier K+ channels was without effect. 6 These data suggest that the relaxant effects of anandamide in sheep coronary arteries are mediated in part via the endothelium and result from the cellular uptake and con-version of anandamide to a vasodilatory prostanoid. This, in turn, causes vasorelaxation, in part, by opening potassium channels.