1. Relaxing factors released by the endothelium and their relative contribution to the endothelium-dependent relaxation produced by bradykinin (BK) in comparison with different vasodilator agents were investigated in human omental resistance arteries. 2. BK produced an endothelium-dependent relaxation of arteries pre-contracted with the thromboxane A2 agonist, U46619. The B2 receptor antagonist, Hoe 140 (0.1, 1 and 10 microM), produced a parallel shift to the right of the concentration-response curve to BK with a pA2 of 7.75. 3. Neither the cyclo-oxygenase inhibitor, indomethacin (10 microM) alone, the nitric oxide synthase inhibitor, N omega-nitro-L-arginine methyl ester (L-NAME, 300 microM) alone, the nitric oxide scavenger, oxyhaemoglobin (Hb, 10 microM) alone, nor the combination of L-NAME plus Hb affected the concentration-response curve to BK. Conversely, the combination of indomethacin with either L-NAME or Hb attenuated but did not abolish the BK-induced relaxation. By contrast, the relaxations produced by the Ca2+ ionophore, calcimycin (A23187), and by the inhibitor of sarcoplasmic reticulum Ca(2+)-ATPase, thapsigargin (THAPS), were abolished in the presence of indomethacin plus L-NAME. Also, the presence of indomethacin plus L-NAME produced contraction of arteries with functional endothelium. 4. The indomethacin plus L-NAME resistant component of BK relaxation was abolished in physiological solution (PSS) containing 40 mM KCl and vice versa. However, in the presence of KCl 40 mM, indomethacin plus L-NAME did not affect the nitric oxide donor, S-N-acetylpenicillamine-induced relaxation. 5. The indomethacin plus L-NAME resistant component of the relaxation to BK was significantly attenuated by the K+ channel blocker tetrabutylammonium (TBA, 1 mM). However, it was not affected by other K+ channel blockers such as apamin (10 microM), 4-aminopyridine (100 microM), glibenclamide (10 microM), tetraethylammonium (10 mM) and charybdotoxin (50 nM). 6. In the presence of indomethacin plus L-NAME, the relaxation produced by BK was not affected by the phospholipase A2 inhibitor, quinacrine (10 microM) or by the inhibitor of cytochrome P450, SKF 525a (10 microM). Another cytochrome P450 inhibitor, clotrimazole (10 microM) which also inhibits K+ channels, inhibited the relaxation to BK. 7. These results show that BK induces endothelium-dependent relaxation in human small omental arteries via multiple mechanisms involving nitric oxide, cyclo-oxygenase derived prostanoid(s) and another factor (probably an endothelium-derived hyperpolarizing factor). They indicate that nitric oxide and cyclo-oxygenase derivative(s) can substitute for each other in producing relaxation and that the third component is not a metabolite of arachidonic acid, formed through the cytochrome P-450 pathway, in these arteries.