The molecular, kinetic, and regulatory characteristics of bovine liver glutamate dehydrogenase have been examined as a function of modification of lysyl residues by trinitrobenzenesulfonate. It is shown that modification of three of the six chains of the monmmer at lysine-428 of the tentative sequence (Smith, E. L., Landon, M, Piszkiewicz, D., Brattin, W. J., Jr., Langley, T. J., and Melamed, M. D., Proc. Nat. Acad. Sci. U. S. 67, 724 (1970)) results in the loss of excess NADH inhibition, an apparent change in the dissociation constant for GTP and ADP when using NADH as coenzyme, some loss in NAD activation, and a dissociation of the polymeric form of the enzyme to the monomer. Further trinitrophenylation, resulting in the modification of lysine-425 up to three groups per six chains, as well as small extents of modification of a number of other lysyl residues, gives rise to an apparently homogeneous polymerized form of the enzyme and it is concluded that the polymerization arises from the modification of lysine-425 or the sum of modification of lysine-428 and -425. Although some activity is lost under these conditions, similar results are obtained under conditions where ligand addition prevents loss of activity. In contrast to native enzyme, the sedimentation behavior of the modified polymerized enzyme, whether active or not, is unaffected by the presence of coenzyme and purine nucleotides. Furthermore, this enzyme appears to have characteristics more closely related to native monomrr than polymer. The results obtained in these experiments are compared to those of others. It is noted that complete kinetic studies as well as identification of reacted residues should accompany chemical modification experiments if correct interpretation of the data is desired, particularly for complex enzyme systems.