Endothelium-dependent relaxations cannot be fully explained by the release of either NO or/and prostacyclin. Another unidentified substance(s) which hyperpolarizes the underlying vascular smooth muscle cells may contribute to endothelium-dependent relaxations, especially in small arteries. It has been termed endothelium-derived hyperpolarizing factor (EDHF). In blood vessels from various species including humans, endothelium-dependent relaxations are partially or totally resistant to inhibitors of NO synthase and cyclooxygenase and are observed without an increase in the intracellular level of cyclic nucleotides in the vascular smooth muscle cells. In some species (canine, porcine and human) nitrovasodilators do not cause hyperpolarization while in other (rat, guinea-pig, rabbit), they evoke glibenclamide-sensitive hyperpolarization, suggesting the involvement of ATP-dependent potassium channels. In contrast, hyperpolarizations caused by EDHF are insensitive to glibenclamide but are inhibited by apamin or the combination of charybdotoxin plus apamin, indicating that NO and EDHF interact with two different targets. The existence of EDHF as a diffusable substance has been demonstrated under bioassay conditions whereby the source of EDHF was either native vascular segments or cultured endothelial cells. The identification of EDHF may allow a better understanding of its physiological and pathophysiological role(s).