G el filtration studies on Bio-Gel P-10 columns of a 50-fold purified porcine duodenal extract revealed a main peak of glucagon-like immunoreactivity (GLI) in the 2,900 mol wt zone and a smaller peak in the 3,500 mol wt zone, the same zone as the pancreatic glucagon marker. Like pancreatic glucagon, samples of 3,500 mol wt material gave essentially identical measurements in radioimmunoassays employing the pancreatic glucagon-specific antiserum 30K and the GLI crossreacting antiserum 78J, whereas the 2,900 mol wt peptide gave 60-fold higher readings in the 78J assay. On disk gel electrophoresis, the 3,500 mol wt fraction, like pancreatic glucagon, migrated at pH 8.3, whereas the 2,900 mol wt peptide remained at the origin; at pH 4.7, the 2,900 mol wt peptide migrated while the 3,500 mol wt immunoreactive peptide and glucagon remained at the origin. Isoelectric focusing revealed the 3,500 mol wt moiety to have an isoelectric point (pI) of 6.2, the same as pancreatic glucagon, whereas the 2,900 mol wt peptide had an pI greater than 10. The glycogenolytic activity of the 3,500 mol wt peptide in the perfused rat liver did not differ significantly from glucagon, and its adenylate cyclase stimulating activity in partially purified liver cell membranes was comparable to that of glucagon; the 2,900 mol wt peptide had less than 20% of these activities. In samples of 3,500 mol wt material subjected to isoelectric focusing, adenylate cyclase-stimulating activity was confirmed to fractions containing 30K immunoreactivity with a pI of 6.2. In samples of 2,900 mol wt material subjected to isoelectric focusing, adenylate cyclase-stimulating activity was confined to fractions containing 78J immunoreactivity with an pI greater than 10. Displacement of [125-I]glucagon from the membranes was limited to these two biologically active fractions. However, the affinity of both pancreatic glucagon and the 3,500 mol wt peptide was an order of magnitude greater than of the 2,900 mol wt peptide. Thus, by all of several biologic, physiocochemical, and immunometric techniques, the 3,500 mol wt gut immunoreactive peptide could not be distinguished from pancreatic glucagon, while the 2,900 mol wt peptide was readily differentiated by all these techniques. "True" A-cells, ultrastructurally indistinguishable from pancreatic A-cells but differing from the A-like cells of the lower bowel, were identified in the gastric fundus of dogs. Their distribution corresponded to that of the 3,500 mol wt immunoreactivity resembling pancreatic glucagon, while the distribution of "A-like cells" in the lower small intestine corresponded to that of GLI.