The water structure has been analyzed for a model of the protein crambin refined against 0.945-x-ray diffraction data. Crystals contain 32% solvent by volume, and 77% of the solvent molecules have been located-i.e., 2 ethanol molecules and 64 water molecules with 10-14 alternate positions. Many water oxygen atoms found form chains between polar groups on the surface of the protein. However, a cluster of pentagonal arrays made up of 16 water molecules sits at a hydrophobic, intermolecular cleft and forms a cap around the methyl group of leucine-18. Several waters in the cluster are hydrogen-bonded directly to the protein. Additional closed circular arrays, which include both protein atoms and other water oxygen atoms, form next to the central cluster. This water array stretches in the b lattice direction between groups of three ionic side chains.Water is the most abundant molecule in living systems and plays an important role in their structure and function. From crystal structures of proteins (1-3), the following picture of tightly bound internal and surface water molecules has emerged (4, 5). First, internal water is found singly (6), in clusters (7), or bound to metal ions (8, 9). It either may stabilize the protein structure by connecting charged or polar groups, or both, or may serve a catalytic function (10). Second, surface water links polar groups at the intra-and intermolecular protein surface; 2-3 times more water hydrogen bonds are made to main-chain -CO groups than to -NH groups (4, 5) in both crystal structures and simulations, reflecting in part the greater capacity of -CO to form hydrogen bonds. Many surface waters bind at turn regions (11). Here, fewer secondary-structure, backbone hydrogen bonds are formed, and the side chains are often hydrophilic (12). Little if any ordered surtace water has been reported at nonpolar side chains, perhaps because such water may be mobile as inferred from computer simulations (13). Finally, less ordered water (about 30% of that in the first shell) is seen in the second shell (14).Water-protein interactions may be important for protein folding (4). Klotz proposed that hydrophobic residues would organize surface water into five-membered ring arrays analogous to the water clathrate (cage) hydrate structures (15). However, protein crystal structures reveal that many hydrophobic groups fold instead into the protein interior away from water, as suggested by Kauzmann (16). Nevertheless, 40-50%o are found on the protein surface in contact with solvent. The folded structure may be a balance between removal of hydrophobic groups from water to form van der Waals attractive contacts and the penalty for burial of polar groups which might not be able to form hydrogen bonds. The arrangement of water at surface hydrophobic sites is not known. If five-membered ring structures were formed (15) and were not hydrogen-bonded to the protein, they might become disordered or vibrate strongly. Such waters would contribute little scattering to an x-ray experiment.Crystals of the h...