The ultimate goal of a QSAR analysis is prediction, which depends on the elaboration of the most appropriate set of molecular descriptors. As such, molecular description is the nucleus of QSAR and in the absence of exhaustive molecular description, rational drug design may be greatly impeded. As previously discussed, computational methods such as quantum mechanics and molecular mechanics provide molecular description at a fundamental level which then enhances the descriptive capability and predictive power of a QSAR analysis. In recognition of these capabilities, semi-empirical molecular orbital methods and molecular mechanics now have been incorporated into or interphased with QSAR programs. Such integrated packages are being successfully used in computer-aided molecular modeling. Computer-aided molecular modeling can provide the three-dimensional structure of a molecule, its chemical and physical characteristics, comparisons of structures of different molecules, and visualization of complexes formed between them. From the foregoing, predictions may be made about how related new molecules may function. Thus, the combination of quantum and/or molecular mechanics and QSAR provides a formidable weapon in the chemist's armamentarium. The molecular modeling approaches are certainly more practical to use than physicochemical methods. They also provide electronic and thermodynamic data that are not available from x-ray crystallographic data. Of course, these techniques are not confined to radiopharmaceutical development and they also could aid in the development of contrast agents for radiography or magnetic resonance imaging. We believe that as computational resources and capabilities increase over the next decade, computer-aided drug design will become a standard procedure in all drug development laboratories.