Attempts to correlate biological activity and chemical structure have had limited success unless the study has confined itself to congeneric series, i.e. sets of analogs in which the chemical variation is sufficiently limited that the basis of molecular comparison is obvious. However, recen+ advances1 in computational chemistry, molecular graphics and problem formulation appear to promise general solutions to the problem of comparison of structurally diverse molecules interacting at the same receptor site. Two major problem classes may be distinguished: those in which the three-dimensional structure of the receptor site has been determined by crystallography, or inferred by model building from known homologous structures, and those in which the receptor structure has not been determined. Garland Marshall and Richard Cramer review the current state of development of each approach with examples that serve to illustrate the potential impact of these advances on our understanding of molecular recognition and development of future, novel therapeutic agents. Selective 5HT1A and adenosine ligands have already been developed using such techniques.