1. The effects of NG-monomethyl-L-arginine, tetrodotoxin and glibenclamide on hypoxia-induced coronary artery relaxation, induced by bubbling Krebs solution with 95% N2 and 5% CO2 instead of 95% O2 and 5% CO2, were assessed by measuring the changes in isometric tension in isolated epicardial coronary artery rings of the rabbit. In addition, the effects of glibenclamide on the relaxation induced by adenosine were investigated. 2. Hypoxia caused a transient relaxation of 38 +/- 3% (P < 0.01) and 17 +/- 2% (P < 0.01) in endothelium-intact or -denuded arteries respectively. NG-monomethyl-L-arginine (30 and 100 microM) inhibited the relaxation in endothelium-intact rings to 31 +/- 2% (P < 0.05) and 16 +/- 2% (P < 0.01) respectively and slightly but significantly attenuated the relaxation in endothelium-denuded rings to 15 +/- 1% and 13 +/- 1% (P < 0.05) respectively. 3. Glibenclamide, a potassium channel inhibitor, did not significantly after the hypoxia-induced relaxation. 4. Incubation with tetrodotoxin (3 and 10 microM) for 30 min reduced the relaxation to 31 +/- 3% (P < 0.05) and 14 +/- 2% (P < 0.01), and 14 +/- 2% (P < 0.05) and 11 +/- 1% (P < 0.05) in endothelium-intact and -denuded rings respectively. However, indomethacin (10 microM), atropine (1 microM), propranolol (10 microM) and phentolamine (10 microM) did not significantly affect the relaxation. 5. Adenosine (1, 10 and 100 MicroM) caused relaxation of 6 +/- 1%, 52 +/-3% and 97 +/-2% respectively in endothelium-denuded rings precontracted with prostaglandin F2alpha (PGF2 alpha, 3 MicroM) and the relaxation was markedly inhibited by 8-phenyltheophylline. Furthermore, glibenclamide (1 and 10 MicroM) reduced the relaxation induced by adenosine (1, 10 and 100 MicroM) to 2 +/-1% (P<0.05), 38 =/-3% (P<0.05) and 85 +/-2%(P<0.05), and 0.6 +/- 0.4% (P<0.05), 27 +/- 4% (P<0.05) and 72 +/- 4% (P<0.01) respectively, in these endothelium-denuded preparations.6. These data suggest that hypoxia-induced relaxation is mediated by the release of nitric oxide rather than by the activation of glibenclamide-sensitive potassium channels in rabbit isolated coronary arteries. A neurogenic mechanism partially modulates the relaxation, possibly by activating non-adrenergic and noncholinergic nerve endings. The inhibition by glibenclamide on adenosine-induced relaxation in isolated coronary arteries may help to explain the fact that glibenclamide inhibits hypoxic coronary relaxation in perfused hearts but not in isolated coronary preparations.