1 Endothelium-derived hyperpolarizing factor (EDHF) has recently been identi®ed as potassium released from endothelial cells into the myo-endothelial space. The present study was designed to test this hypothesis. 2 In rat small mesenteric arteries, mounted in a wire myograph, relaxation to acetylcholine or potassium was not signi®cantly changed following incubation with oxadiazolo-quinoxalin-1-one (ODQ, 4 mM) and indomethacin (10 mM, n=9). 3 Maximal relaxations to acetylcholine occurred in all arteries, were maintained and were signi®cantly greater (P50.01, n=9) than the transient relaxations to potassium, which only occurred in 30 ± 40% of vessels. 4 Removal of the vascular endothelium abolished relaxant responses both to potassium and acetylcholine (P50.005, n=9). 5 Compared with responses in 5.5 mM potassium PSS, relaxation responses to added potassium in arteries maintained in 1.5 mM potassium PSS were more marked and were not dependent on the presence of an intact endothelium (n=8). 6 Incubation with BaCl 2 (50 mM) signi®cantly inhibited the maximal relaxant response to potassium in the presence of an intact endothelium in 5.5 mM potassium PSS (P50.05, n=4), but had no eect on relaxation of de-endothelialized preparations in 1.5 mM potassium PSS (n=5). 7 Treatment with ouabain (0.1 mM) abolished the relaxant response to potassium in 1.5 mM potassium PSS (P50.001, n=9), but only partly inhibited the maximal relaxant response to acetylcholine in 5.5 mM potassium PSS (P50.01, n=5). 8 These data show that at physiological concentrations of potassium an intact endothelium is necessary for potassium-induced relaxation in rat mesenteric arteries. Furthermore, the response to potassium is clearly dierent to that from acetylcholine, indicating that potassium does not mimic EDHF released by acetylcholine in these arteries.