A common three-dimensional receptor model has been formulated for six different classes of rat liver dihydrofolate reductase inhibits using the distance geometry approach. Altogether, 62 molecules of five different classes were used to generate the receptor model, which has 11 attractive site points and 5 repulsive ones. It gave a fit having a correlation coefficient of 0.949 and root mean square (rms) deviation of 0.527. The attractive site points of the model closely correspond to the one we reported earlier. The model successfully predicted the biological data of 33 molecules of 5 different classes, one molecule of which was a member of a new class not included in the original data set. Guidelines are put forth for the synthesis of improved inhibitors.