To achieve optimal binding, enzyme inhibitors, receptor antagonists and agonists, or DNA/RNA binders must exhibit optimal molecular recognition with the macromolecular biological target structure, a principle that has been the basis for the development of computational drug design methods. Predicting the correct binding mode of a ligand in a protein invokes the prior positioning of the ligand by a search engine that ensures an efficient and unbiased sampling. However, most of the developed models do not account for conformational changes in the protein upon binding or presence of key water molecules. As a consequence, when activity is correlated to induced fit effects or bridging water molecules, usual approaches often lead to misleading results. The binding mode of interest must next be identified and the binding affinity predicted. For this purpose, a variety of scoring functions have been developed. However, the available scoring functions rank (to some extent) compounds according to their biological activity but their predictiveness still relies heavily on the target under study. To account for side-chain or backbone adjustments, presence of water molecules and to address the scoring function predictiveness, we have developed new strategies [1,2], now implemented in the current version of FITTED [3]. The development of FITTED and its application to a variety of enzymes, receptors, and RNA aptamers will be presented.