The basic idea behind ligand-based approaches is that the analysis of sets of molecules with experimentally determined activities can highlight those chemical features responsible for the activity changes. Historically, such approaches have been devised before structure-based methods. Nowadays, despite the ever increasing availability of experimentally determined structures, ligand-based approaches still play a major role in drug design either alone or in conjunction with structure-based efforts. This manuscript aims to provide a general overview of the main computational approaches in ligand-based drug discovery, particularly 3D QSAR methods, along with relevant references to the literature.