The optimisation of ligand-macromolecule interactions is fundamental to the design of therapeutic agents. The GRID method is a procedure for determining energetically favourable ligand binding sites on molecules of known structure using an empirical energy potential. In this thesis, it has been extended, tested, and then applied to the design of anti-influenza agents. In the GRID method, the energy of a hydrogen-bond is determined by a function which is dependent on the length of the hydrogen-bond, its orientation at the hydrogen-bond donor and acceptor atoms, and the chemical nature of these atoms. This function has been formulated in order to reproduce experimental observations of hydrogen-bond geometries. The reorientation of hydrogen atoms and lone-pair orbitals on the formation of hydrogen-bonds is calculated analytically. The experimentally observed water structures of crystals of four biological molecules have been used as model systems for testing the GRID method. It has been shown that the location of well-ordered waters can be predicted accurately. The ability of the GRID method to assist in the assignment of water sites during crystallographic refinement has been demonstrated. It has also been shown that waters in the active site of an enzyme may be both stabilized and displaced by a bound substrate. Ligands have been designed to block the highly conserved host cell receptor site of the influenza virus haemagglutinin in order to prevent the attachment of the virus to the host cells. The protein was mapped energetically by program GRID and specific ligand binding sites were identified. Ligands, which exploited these binding sites, were then designed using computer graphics and energy minimization techniques. Some of the designed ligands were peptides and these were synthesised and assayed. Preliminary results indicate that they may possess anti-influenza activity.