Several new 3 beta-(4'-substituted phenyl)-, 3-beta-(3'-substituted phenyl)-, and 3 beta-(3',4'-disubstituted phenyl)tropane-2 beta-carboxylic acid methyl esters were prepared and assayed for inhibition of [3H]WIN 35,428 binding to the dopamine transporter. The 3 beta-(3',4'-dichloro) and 3 beta-(4'-chloro-3'-methyl) analogues (2w and 2y; RTI-111 and RTI-112, respectively) with IC50 values of 0.79 and 0.81 nM showed the highest affinity. The contributions of quantitative structure-activity relationship (QSAR) models derived from the classical and comparative molecular field analysis (CoMFA) approaches to rational drug design were examined. CoMFA models were derived using steric and electrostatic potentials with SYBYL default values while the classical models were derived from pi and MR parameters. Using a 12-compound training set, both models were used for predicting the binding affinity of compounds both inside and outside the training set. The CoMFA study provided new insight into the steric and electrostatic factors influencing binding to the DA transporter and provided additional support for our original finding that CoMFA is useful in predicting and designing new compounds for study. The classical QSAR models, which were easier to obtain, suggest that the distribution property (pi) of the compounds is an important factor. Overall, the SAR, CoMFA, and conventional QSAR studies elaborated some features of the cocaine binding site pharmacophore and provided useful predictive information.