Subdiabetogenic doses of alloxan injected into the hepatic portal vein of rats abolished glucagon-induced inhibition of feeding (glucagon satiety) both in daytime tests using a palatable food and in nighttime tests using their standard pelleted diet. In contrast, inhibition of food intake by cholecystokinin and epinephrine and stimulation of feeding by 2-deoxy-D-glucose were not impaired by alloxan. Alloxan-induced deficits in glucagon satiety did not appear to result from generalized hepatocellular necrosis, because satiety deficits outlasted histological signs of toxicity and because furosemide, which produced a similar degree of hepatotoxicity, did not impair glucagon satiety. In addition, alloxan's effects were not associated with impaired glycogen storage or mobilization. Recovery of glucagon satiety occurred in some animals but not until 3-6 mo after alloxan. The degree of recovery was inversely related to alloxan dose. Our results indicate that, when administered into the hepatic portal vein, alloxan may be a relatively specific toxin for cells involved in the mediation of glucagon satiety. The specificity of the deficit and the time course of recovery suggest that the alloxan-sensitive cells may be hepatic vagal neurons.