1. The rat isolated superior mesenteric bed, perfused with Krebs-Henseleit solution containing 10 microM indomethacin and precontracted with 100 microM methoxamine, was used to study the vasorelaxation produced by some fragments of endothelin-1, by two alanyl-substituted analogues, and by human and porcine proendothelins. The systemic cardiovascular effects of some of these peptides were also studied in anaesthetized rats. 2. Endothelin (16-21) was an endothelium-dependent vasodilator about 10 times less potent than acetylcholine and its amide was about 500 times less potent than the free acid. Human proendothelin-1 and porcine proendothelin-1 also caused endothelium-dependent relaxations but neither [Ala3,11]endothelin-1 nor [Ala1,3,11,15]endothelin-1 produced significant reductions in the methoxamine-induced tone. Sodium nitroprusside was approximately 200 times less potent than acetylcholine in the presence of the endothelium and was the only vasorelaxant to be active after destruction of the endothelium by perfusion with 0.3% CHAPS; in the absence of the endothelium it was 3.7 times more potent as a vasodilator than in its presence. 3. Porcine proendothelin-1 had weak contractile activity in the isolated mesenteric bed but porcine endothelin (22-39), endothelin (16-21) and endothelin (16-21) amide did not have vasoconstrictor actions. 4. Endothelin-3, [Ala3,11]endothelin-1, [Ala1,3,11,15]endothelin-1 and endothelin (16-21) all had systemic blood pressure effects in the anaesthetized rat. Although endothelin (16-21) did not cause vasoconstriction in vitro, it increased mean arterial pressure in vivo but was at least 100 times less potent than endothelin-3. Despite causing no vasorelaxation in vitro, [Ala1,3,11,15]endothelin-1 and [Ala3,11]endothelin-1 induced short-lived systemic depressor responses which were followed by pressor responses. Endothelin-3 was more potent as a systemic depressor agent than as a pressor agonist and, as a vasodepressor agent, it was 3-4 times more potent than either of the alanyl-substituted analogues. Endothelin-3, [Ala3,11]endothelin-1 and [Ala1,3,11,15]endothelin-1 were equipotent vasopressor agents. Porcine endothelin (22-39) had no systemic blood pressure effects. 5. This study shows directly that the presence of both disulphide bonds is required in the endothelins for them to be able to cause endothelium-dependent relaxation in the mesenteric vascular bed and that this response is mediated by different receptors from those causing contraction of the vascular smooth muscle in the mesentery. Vasorelaxation caused by endothelin (16-21) and its amide suggests that there is a receptor on the endothelium similar to one reported in the guinea-pig trachea. Finally, the ability of endothelin peptides to cause systemic vasodepressor responses is not related to their ability to cause endothelium-dependent relaxation, at least in the mesenteric circulation, and, for endothelin (16-21), its systemic pressor response is not likely to be the result of vasoconstriction in the mesenteric bed.