The results reported in the literature regarding glucagonaemia in genetically obese fa/fa rats are conflicting: normal, increased or decreased plasma glucagon levels have been reported. Due to the existence of several molecules endowed with glucagon-like immunoreactivity, it was thought that the conflicting data could be related to the degree of specificity of the different glucagon antibodies. Three antibodies that all qualified as being specific for pancreatic glucagon were used. It was found that, depending on the antibody, absolute values of basal glucagonaemia or arginine-induced glucagon output varied quantitatively and qualitatively in both lean and obese rats. When non-extracted basal or stimulated plasma samples were passed on a G-50 Sephadex column, glucagon-like immunoreactivity was present over a wide range of molecular weights, indicating the presence of non-pancreatic glucagon molecules. When an ethanol extraction was used, the fractions eluting from the G-50 Sephadex column contained only pancreatic glucagon immunoreactivity. It is concluded that ethanol extraction is necessary for the measurement of the 3500 daltons glucagon. Using this methodology it was found that: (1) basal glucagonaemia was low but identical in the two groups of rats; (2) arginine-induced glucagon secretion was greater in obese than in lean animals; (3) glucagonaemia was decreased by glucose administration in lean but not in obese rats. It is concluded that there are, in obese animals, dysfunctions of glucagon output that may play a role in their abnormal glucose tolerance.