With the number of protein-ligand complexes available in the Protein Data Bank constantly growing, structure-based approaches to drug design and screening have become increasingly important. Alongside this explosion of structural information, a number of molecular docking methods have been developed over the last years with the aim of maximally exploiting all available structural and chemical information that can be derived from proteins, from ligands, and from protein-ligand complexes. In this respect, the term guided docking is introduced to refer to docking approaches that incorporate some degree of chemical information to actively guide the orientation of the ligand into the binding site. To reflect the focus on the use of chemical information, a classification scheme for guided docking approaches is proposed. In general terms, guided docking approaches can be divided into indirect and direct approaches. Indirect approaches incorporate chemical information implicitly, having an effect on scoring but not on orienting the ligand during sampling. In contrast, direct approaches incorporate chemical information explicitly, thus actively guiding the orientation of the ligand during sampling. Direct approaches can be further divided into protein-based, mapping-based, and ligandbased approaches to reflect the source used to derive the features capturing the chemical information inside the protein cavity. Within each category, a representative list of docking approaches is discussed. In view of the limitations of current scoring functions, it was generally found that making optimal use of chemical information represents an efficient knowledge-based strategy for improving binding affinity estimations, ligand binding-mode predictions, and virtual screening enrichments obtained from protein-ligand docking.