The concept of structure-based drug design is based upon an in-depth understanding of the principles of molecular recognition. Despite our lack of a thorough comprehension of these principles, the wealth of protein structures available opens up unprecedented possibilities for new insights from the analysis of these data. Unravelling universal rules of molecular recognition is certainly one of the most appealing goals. But our knowledge is enhanced also when studying the specific determinants that characterise single targets or target families only, and the factors governing and discriminating their recognition properties.Here, we illustrate how the structure-based design process can benefit from the consequent incorporation of database query tools. We discuss representative examples to address issues such as protein flexibility, water molecules in binding pockets, and ligand specificity as some of the most critical aspects of drug design. All studies are carried out using our database system Relibase. We also show the application of Relibase in searching for preferred geometrical patterns between interacting molecular fragments.