Features of your target structure
Depending on the structural features of the target model, you may need to take special steps to prepare for docking. Many of the following points are relevant no matter which docking program you may use. Comments that are specific to a particular version or particular kind of docking program are noted.
Covalent inhibitor bound
If the target model you wish to dock to has a covalent inhibitor bound, your first step is obviously to remove the inhibitor. You should also keep in mind, however, that the covalently bound ligand may have distorted the active site in ways that a non-covalently bound ligand would not. If you have access to a structure with a non-covalently bound ligand, examine the differences - it may well be preferable to use the non-covalently bound ligand form of the target.
Non-covalent inhibitor bound
Dimer or multimer
Use PQS.ebi.ac.uk to check multimeric state. Is the binding site near a protein-protein interface? If not, use the monomer and ignore the multimeric state. The additional complexities of including a larger structure do not outweigh the more accurate electrostatic potential, in general. However, if the binding site is within say 6 A of the protein-protein interface, then at least a fragment of the second protein must be used in the model. From 6-10 A it is a judgement call, but is probably advisable to include Watch out also for ligand binding that significantly affects the relative orientation of the two proteins.