Knowledge of the spatial structure of complexes formed by cellular proteins and membrane receptors with their espective ligands is an important step towards understanding the mechanisms of their functioning. Rational drug design and the earch for new therapeutically active compounds also require structural information on the interaction of prototypic drugs with the target protein. The present review briefly describes the main computational methods of molecular docking that are used to predict the conformation of a ligand bound to the active center of a protein. Approaches enabling an increase of the precision and efficiency of the currently used docking algorithms are exemplified by the recent projects of the Laboratory of Biomolecular Modeling of IBCh RAS. Special attention is paid to hydrophilic and hydrophobic interactions, as well as to the stacking henomena that account for the molecular recognition of specific ligand fragments. These types of contacts are often inadequately described by the algorithms of the estimation of the intermo- lecular interaction energy of the existing docking programs (scoring functions), his ultimately leading toerroneous predictions of the three-dimensional structure of complexes. Therefore, a thorough consideration of hese interactions is one of the most important tasks of molecular modeling. © Pleiades Publishing, Ltd., 2010.