The vasodilating potencies and mechanism of action of KRN2391 [N-cyano-N'-(2-nitroxyethyl)-3-pyridinecarboximidamide monomethanesulfonate] and its metabolite Ki3315 [N-cyano-N'-(2-hydroxyethyl)-3-pyridinecarboximidamide] were investigated in rat isolated aortae and in pithed rats. In experiments using rat isolated aortae, Ki3315 (10(-5)-3 x 10(-4) M) and KRN2391 (10(-8)-10(-4) M) produced concentration-dependent relaxations. However, the vasodilating potency of Ki3315 was about 1000-fold lower than that of KRN2391. The relaxant effect of KRN2391 was antagonized by both methylene blue and glibenclamide, whereas that of Ki3315 was only antagonized by glibenclamide. In experiments using pithed rats, cumulative administration of Ki3315 (0.3-30 mg/kg, i.v.) and KRN2391 (3-300 micrograms/kg, i.v.) produced a dose-dependent decrease of diastolic blood pressure. The vasodepressor effect of Ki3315 was weaker than that of KRN2391. In rats given glibenclamide (20 mg/kg, i.v.), the vasodepressor curves for Ki3315 and KRN2391 were shifted to the right. The ED50mmHg values increased about 5-fold for KRN2391 and 8.2-fold for Ki3315 in glibenclamide-treated rats. These results suggest that Ki3315 and KRN2391 show vasorelaxant effects in vitro and in vivo, but in this respect Ki3315 is weaker than KRN2391. These effects of Ki3315 are predominantly due to a K channel-opening action, while those of KRN2391 involve a nitrate action in addition to a K channel-opening action.