1. The survival of the endothelium-derived relaxing factor (EDRF) released by acetylcholine (ACh) in the rat isolated perfused mesenteric vascular bed was compared to that of exogenously applied nitric oxide (NO) by direct bioassay of the effluent from the mesenteric bed on rabbit aortic strips (RbAs). 2. NO (0.1-10 nmol) produced dose-related vasodilatations in the mesenteric vascular bed and also survived in the effluent in concentrations sufficient to provoke relaxations of the RbAs. ACh (1.7 pmol-300 nmol) caused dose-related vasodilatations in the mesenteric bed but no EDRF was detected in the effluent. This was true even when ACh provoked much larger vasodilatations in the mesenteric bed than those to doses of NO that survived passage through the bed. 3. Removal of the endothelial cells in the mesenteric vascular bed abolished vasodilatations in response to ACh but did not significantly affect those to NO. Survival of NO through the mesenteric bed was not increased. 4. Superoxide dismutase (SOD, 10 u ml-1) did not significantly affect the vasodilator effects of either ACh or NO. It did increase the survival of NO through the mesenteric vascular bed but native EDRF was still not detected in the effluent. 5. The absence of bioassayable EDRF in the effluent following ACh-induced vasodilatations might be explained by the volume into which EDRF is released abluminally being much smaller than that into which it is released luminally.