Water bound to proteins is a ubiquitous and integral component of the protein surface, as observed by X-ray crystallography. A substantial proportion of surface-bound water is required for protein activity, and remains even under unfavorable conditions such as repeated rinsing in organic solvent. Bound water also plays essential and diverse roles in protein-ligand interactions; for instance, water is required for the recognition of operator DNA by trp repressor, and is essential not only for inhibitor binding by trypsin, but also for trypsin catalysis. This proposal addresses several outstanding questions concerning protein-water and proteinwater-ligand interactions. What is the chemistry, beyond hydrogen bonding, that determines the favorability of water binding sites and their conservation between structures? How can this knowledge be to used to assess the favorability of sites and to predict hydration? In particular, given the uncomplexed structure of a protein or ligand, can the bound water sites mediating the protein-ligand interaction be predicted? Using the same techniques designed to answer these questions, advanced software for modeling bound water will be developed and distributed. Key applications of this software will be to fill two major gaps in current solvent modeling, by providing a knowledge-based, uniform approach to water site assignment for crystallography, and by identifying active-site hydration appropriate to include in computer-guided ligand design and docking.