1. The effects of vasopressin and deamino-8-D-arginine vasopressin (DDAVP, desmopressin) were studied in artery rings (0.8-1 mm in external diameter) obtained from portions of human omentum during the course of abdominal operations (27 patients). 2. In arterial rings under resting tension, vasopressin produced concentration-dependent, endothelium-independent contractions with an EC50 of 0.59 +/- 0.12 nM. The V1 antagonist d(CH2)5Tyr(Me)AVP (1 microM) and the mixed V1-V2 antagonist desGly-d(CH2)5D-Tyr(Et)ValAVP (0.01 microM) displaced the control curve to vasopressin to the right in a parallel manner without differences in the maximal responses. In the presence of indomethacin (1 microM) the contractile response to vasopressin was significantly increased (P < 0.01). 3. In precontracted arterial rings, previously treated with the V1 antagonist, d(CH2)5Tyr(Me)AVP (1 microM), vasopressin produced endothelium-dependent relaxation. This relaxation was reduced significantly (P < 0.05) by indomethacin (1 microM) and unaffected by the V1-V2 receptor antagonist desGly-d(CH2)5D-Tyr(Et)ValAVP (1 microM) or by NG-nitro-L-arginine methyl ester (L-NAME, 0.1 mM). 4. The selective V2 receptor agonist, DDAVP, caused endothelium-independent, concentration-dependent relaxations in precontracted arterial rings that were inhibited by the mixed V1-V2 receptor antagonist, but not by the V1 receptor antagonist or by pretreatment with indomethacin or L-NAME. 5. Results from this study suggest that vasopressin is primarily a constrictor of human mesenteric arteries by V1 receptor stimulation; vasopressin causes dilatation only during V1 receptor blockade. The relaxation appears to be mediated by the release of vasodilator prostaglandins from the endothelial cell layer and is independent of V2 receptor stimulation or release of nitric oxide. In contrast, the relaxation induced by DDAVP is largely dependent on stimulation of V2 receptors.