The functions of the Trp-25 and Met-27 residues and the free carboxy terminus of glucagon have been debated for many years. Despite some semi-synthetic data to the contrary, comparison of the glucagon sequence with the other 5 members of this family of peptides, all of them amides and particularly growth hormone-releasing factor(1-29) amide and its recently described analogues, suggests that alterations to these positions should be quite well tolerated in terms of biological activity. To test this prediction, [Phe-25,Leu-27]-glucagon amide was synthesized in high yield and was found to actually have superior glycogenolytic activity (196%) to glucagon in the rat. Replacement of Gly-4 by D-Phe, which has been shown to give much enhanced glycogenolytic activity than glucagon itself, also increased the activity of [D-Phe-4,Phe-25,Leu-27]-glucagon amide (518%). The L-Phe-4-analogue, [Phe-4,25,Leu-27]-glucagon amide, in contrast, was 20 times less active (30%), strongly suggesting the presence of a beta-bend in this N-terminal region of glucagon. This was supported by Chou-Fasman structural predictions which indicate extensive folding in the 1-15 region. Indeed, additional conformational restriction by substitution of D-Ser in position 2 of glucagon also increased activity to 226%. [D-Gln-3]-glucagon was slightly less active (74%) than glucagon. Chou-Fasman calculations on glucagon were compared to similar treatments of the VIP, secretin, PHI, and GRF(1-29) sequences.