Abstract. The three-dimensional-quantitative structure activity relationship (3D-QSAR) studies were performed on a series of direct factor Xa (FXa) inhibitors using AutoGPA-based modeling method in this paper. A training set of 38 molecules and a test set containing 10 molecules were used to build the 3D-QSAR model and validate the derived model, respectively. The developed model with correlation coefficients (r 2 ) of 0.8564 and cross-validated correlation coefficients (q 2 ) of 0.6721 were validated by an external test set of 10 molecules with predicted correlation coefficient (r pred 2 ) of 0.6077. Docking study of FXa inhibitors and FXa active site was performed to check the induced pharmacophore query and comparative molecular field analysis (CoMFA) contour maps using MOE2012.10. It was proved to be coincidence with the interaction information between ligand and FXa active site and was rendered to provide a useful tool to improve FXa inhibitors.