Dipeptidyl peptidase IV (DPP-IV) is a valid drug target for type-2 diabetes and DPP-IV inhibitors have been proven to efficiently improve glucose tolerance. In our study, 3D pharmacophore models were generated using a training set of 22 DPP-IV inhibitors. The best model consisted of important chemical features and mapped well into the active site of DPP-IV. The model gave high correlation coefficients of 0.97 and 0.84 for the training set and the test set, respectively, showing its good predictive ability for biological activity. Furthermore, the pharmacophore model demonstrated the capability to retrieve inhibitors from database with a high enrichment factor of 42.58. All results suggest that the model provides a useful tool for designing novel DPP-IV inhibitors.