1. In guinea-pig internal carotid arteries with an intact endothelium, acetylcholine (10 microM) and levcromakalim (10 microM) each hyperpolarized the smooth muscle whereas a 5 mM elevation of extracellular K(+) was without effect. 2. Incubation of the carotid artery with the gap junction inhibitors carbenoxolone (100 microM) or gap 27 (500 microM) essentially abolished the hyperpolarization to acetylcholine but it was without effect on that to levcromakalim. Carbenoxolone had no effect on the acetylcholine-induced endothelial cell hyperpolarization but inhibited the smooth muscle hyperpolarization induced by the endothelial cell K(+) channel opener, 1-ethyl-2-benzimidazolinone (600 microM). 3. In rat hepatic and mesenteric arteries with endothelium, carbenoxolone (100 or 500 microM) depolarized the smooth muscle but did not modify hyperpolarizations induced by KCl or levcromakalim. In the mesenteric (but not the hepatic) artery, the acetylcholine-induced hyperpolarization was inhibited by carbenoxolone. 4. Phenylephrine (1 microM) depolarized the smooth muscle cells of intact hepatic and mesenteric arteries, an effect enhanced by carbenoxolone. Gap 27 did not have a depolarizing action. In the presence of phenylephrine, acetylcholine-induced hyperpolarization of both hepatic and mesenteric artery myocytes was partially inhibited by each of the gap junction inhibitors. 5. Collectively, the data suggest that gap junctions play some role in the EDHF (endothelium-derived hyperpolarizing factor) response in rat hepatic and mesenteric arteries. However, in the guinea-pig internal carotid artery, electrotonic propagation of endothelial cell hyperpolarizations via gap junctions may be the sole mechanism underlying the response previously attributed to EDHF.