Computational searches for novel ligands for a given protein binding site have become ubiquitous in the pharmaceutical industry, and are potentially equally useful in helping identify small-molecule tools for biology. Here we describe the steps needed to carry out a high-throughput docking (HTD) or three-dimensional (3D) pharmacophore virtual screen starting with a model of the target protein's structure. The advice given is, in most cases, software independent but some tips are provided which apply only to certain popular programs. Useful work can be carried out using free programs on a modest workstation. Of course, any resultant "hits" remain in the virtual world until they are experimentally tested.