Evaluation of the three-dimensional (3D) similarity in a set of chemical structures is a key aspect of molecular modeling. Described here are three new approaches to 3D similarity analysis, each based on interpoint distance comparisons. The approaches are embodied in three computer programs, DISCO, FAMILY, and COMPAIR. Recent applications of these programs discussed are a pharmacophore mapping study for a set of platelet-activating factor (PAF) antagonists, grouping compounds identified from a 3D database search for potential PAF antagonists and comparing search hits to a reference antagonist or to a model of the receptor binding site. The methods and algorithms used in these approaches are briefly described, followed by a discussion of the three recent applications.