Two series of mutant polypeptides of the type I, 37-residue winter flounder "antifreeze" protein have been synthesized and analyzed by nanoliter osmometry, the "ice hemisphere" test, measurement of ice growth hysteresis and circular dichroism (CD) spectroscopy. In series 1 peptides the central two threonines and all four threonines of the native protein were mutated to serine. In series 2 peptides two additional salt bridges (K7, E11 and K29, E33) were incorporated, and all four threonine residues in this sequence were mutated simultaneously to each of serine, valine, alanine, and glycine, respectively. The CD studies showed that all mutants are 100% helical in structure at low temperature, except for the glycine derivative which was estimated to be 70% alpha-helical. Dilute solutions of serine-substituted series 1 peptides showed no detectable, nonbasal faceting, or hysteresis behavior, indicating either no or extremely weak interaction with ice. The analogous serine-substituted mutant in series 2, as well as the glycine derivative, displayed unfaceted growth and showed no hysteresis. Hysteresis values, ice growth patterns, and the helicity measurements showed that the additional salt bridges present in series 2 peptides do not alter significantly the properties of the protein. The valine-substituted mutant gave a distinct etching pattern in which polypeptide accumulates on the {2 0 (2) over bar 1}plane of ice 1h, and exhibited thermal hysteresis comparable to that of the native protein. In the case of the alanine-substituted mutant, reduced hysteresis behavior was measured, together with a distinct etch pattern in the ice hemisphere test. These combined results show that existing hypotheses for the action of native winter flounder peptide are incorrect; these hypotheses include models in which the -OH groups on four threonine side chains, equally spaced 11 residues apart on the 37-residue native polypeptide, are responsible for "binding" of the molecule to the ice/water interface. The antifreeze activity of the valine- and alanine-substituted mutants indicate a significant contribution to the mechanism of ice growth inhibition by type I antifreeze proteins from the hydrophobic methyl group in threonine and valine. Arguments against the importance of the role of hydrogen-bonding are summarized, and alternate ice growth inhibition mechanisms that include hydrophobic interactions are discussed.