Quantitative structure--activity relationship (QSAR) paradigm, using genetic function approximation (GFA) technique was used to examine the correlations between the calculated physicochemical descriptors and the in vitro activities (3'-processing and 3'-strand transfer inhibition) of a series of human immunodeficiency virus type 1 (HIV-1) integrase inhibitors. Depending on the chemical structure, all molecules were divided into two classes---catechols and noncatechols. Eighty-one molecules were used in the present study and they were divided into training set and test set. The training set in each class consisted of 35 molecules and QSAR models were generated separately for both catechols and noncatechols. Equations were evaluated using internal as well as external test set predictions. Models generated for catechols show that electronic, shape related, and thermodynamic parameters are important whereas for noncatechols, spatial, structural, and thermodynamic properties play an important role for the activity.