Reactive oxygen species (ROS) are thought to play a role in EDHF, but endothelial sKCa and IKCa channels opening in response to increased endothelial [Ca2+]i, is also known to be a requirement. We have examined the vasodilatation response to carbachol in the isolated perfused cremaster muscle preparation. The iliac artery of a freshly killed (Schedule 1) Wistar rat was cannulated orthogradely and branches that did not lead to the cremaster muscle were ligated. The cremaster microcirculation was flushed with St. Thomas’ cardioplegic solution {in mmol.l-1: NaCl (110); KCl (7.9); MgCl2.6H20 (34); CaCl2 (1); and Hepes (11)} containing heparin (300 IU.ml-1). The skin over the scrotum was removed to reveal the cremasteric sack. The cremaster was cut to isolate it from the enclosed testes, care being taken not to damage any major vessel. The muscle was spread and pinned over a transparent Sylgard (Dow Corning) support and superfused with a Krebs-bicarbonate buffer solution maintained at 37°C at 1-2 ml.min-1. The perfusate was changed to a buffer containing albumin (10 mg.ml-1) with added FITC-albumin (5 mg.ml-1). The cremaster muscle arterioles and venules (pre-phenylephrine diameter 47 ± 4.0 µm (MEAN ± SEM; n = 15 and 63.2 ± 6.7; n = 12, respectively) dose-dependently dilated to carbachol (artery 68 ± 0.6%, vein 84 ± 2.9%; p < 0.001 paired ‘t’ test for both) which was significantly reduced by L-NAME and indomethacin (artery 48 ± 1.3, vein 70 ± 1.3%). The effect of these inhibitors in reducing the carbachol-induced dilatation was significantly greater in arterioles than in the paired venules (p < 0.05, paired ‘t’ test), which may be accounted for by the differences in the smooth muscle layer thickness. The dilatation was virtually blocked by apamin and charybdotoxin (artery 6 ± 2.0%, vein 12 ± 1.5%, n = 4) and very much reduced by scavenging ROS with SOD and catalase (artery 13 ± 2.9%, vein 15 ± 3.9%, all n = 4).