The secondary structure of antithrombin III was investigated using the predictive scheme based on the amino acid sequence (Fasman, G. D., Chou, P. Y., and Adler, A. J. (1976) Biophys. J. 16, 1201-1238. The relative composition of the conformational regions is: 31% alpha-helix, 16% beta-sheet, 9% beta-turns, and the rest of the molecule in unordered conformation. When the alpha-helical regions are carefully delineated, 44% are predicted to be unstable helices because of unfavorable electrostatic interactions. Circular dichroism studies showed that the disruption of the first unfolding domain in antithrombin III by low concentration of guanidinium chloride (Villanueva, G. B., and Allen, N. (1983) J. Biol. Chem. 258, 14048-14053) was accompanied by a 42% decrease in the alpha-helical content. This suggested that the first unfolding domain demonstrated to be essential for heparin binding (Villanueva, G. B., and Allen, N. (1983) J. Biol. Chem. 258, 11010-11013) corresponds to the predicted unstable helical region. Model-binding studies of the unstable helical segment which contain tryptophan and lysine residues reveal that Lys-282, Lys-286, and Lys-289 are aligned on the same side of the alpha-helix. A model of the antithrombin III-binding octasaccharide (Lindahl, U., Backstrom, G., Thunberg, L., and Leder, I. G. (1980) Proc. Natl. Acad. Sci. U.S.A. 77, 6551-6555) was also constructed based on available NMR and x-ray data of the repeating units in heparin. These two models can be positioned such that the three lysine residues in the alpha-helix can be matched for maximal interaction with the three sulfate groups in the octasaccharide demonstrated as essential for binding antithrombin III. Based on these models, it is proposed that the unstable helical segment composed of residues 281-292 is the heparin-binding site in antithrombin III.