Glucogon immunoreactivity (IRG) was measured in plasma of duodenopancreatectomized subjects with a nonspecific (K-4023) and a specific (30-K) glucagon antiserum. After an overnight fast, plasma IRG (K-4023) was significantly (P < 0.05) higher in the subjects without pancreas, averaging 782+/-79 (SEM) pgeq/ml, than in the controls (482+/-80 pgeq/ml). IRG (30-K) of 162+/-68 pg/ml did not change during an infusion of arginine (450 mg/kg per 40 min). Insulin deprivation during 3 d in one patient did not restore the IRG response to arginine as reported in depancreatized dogs.Bio-Gel P-30 column chromatography revealed that virtually all IRG (30-K) measured in whole plasma was of different molecular weight than glucagon, and primarily of a mol wt >/= 40,000. Intravenous arginine did not significantly alter the chromatographic pattern of these plasmas. Thus, as postulated by others, duodeno-pancreatectomized humans have virtually no circulating 3,500-dalton glucagon. Hence, the presence of 3,500-dalton glucagon in plasma is not a condition for the diabetic state. It might, nevertheless, when present in normal or excessive amounts, worsen the metabolic state of diabetic patients. Among 14 amino acids measured in plasma of these patients, the concentrations of alanine, serine, ornithine, and arginine were significantly (P < 0.05) elevated to approximately twice that of normal: alanine and serine are both substrates for gluconeogenesis, whereas ornithine and arginine are involved in the formation of urea, the second product of hepatic gluconeogenesis. As the concentrations of branched chain amino acids were not grossly altered, it is hypothesized that this amino acid pattern is a consequence of glucagon deficiency rather than secondary to the diabetic state of these patients.