Thirteen years after its discovery, there is still controversy over the chemical identity of endothelium-derived relaxing factor (EDRF). Although pharmacological and chemical evidence indicates that EDRF is nitric oxide, other candidates, including S-nitrosocysteine, dinitrosyl-iron-cysteine complex, nitroxyl and hydroxylamine, have been proposed to account for the vasorelaxant properties of EDRF. Such diverse compounds should differ in their stability and in reactivity with oxyhaemoglobin and with redox-active nucleophiles such as thiols. Here we use a bioassay to compare the pharmacodynamic profiles of these and other compounds with those of nitric oxide and EDRF. We find that some S-nitrosothiols, dinitrosyl-iron-cysteine complex, sodium nitroxyl and hydroxylamine can be eliminated as candidates as they are more stable than EDRF and less susceptible to inhibition by oxyhaemoglobin. Co-infusion of cysteine revealed major differences between the remaining candidates because it reduced the effect of authentic nitric oxide and EDRF on the bioassay tissues but enhanced the survival of S-nitrosocysteine and S-nitrosocysteamine. Our results further support the evidence that EDRF, the pharmacological entity described by Furchgott and Zawadzki, is nitric oxide.