Novel potential human immunodeficiency virus (HIV) protease inhibitors were designed by a combination of nelfinavir and amprenavir motifs. The designed compounds were prepared by a facile synthetic route and their stereochemistry was further confirmed by a stereospecific synthesis from commercially available (S)-2-oxiranylmethyl m-nitrobenzenesulfonate. All compounds were tested for their ability in inhibiting HIV type 1 protease activity with the published method of reference 19. Derivatives 1a--u exhibited moderate to significant inhibitory activities in preliminary bioassay. The best compound 1a has IC50 value of 0.02 microM, comparable to that of amprenavir. A docking study on compounds 1a--u was performed using the published X-ray crystal structure of HIV type 1 protease, all compounds bound to the HIV type 1 protease in an extended conformation and the scaffoldings of the binding conformations could be aligned quite well. Comparative molecular field analysis (CoMFA) study was performed to explore the specific contributions of electrostatic and steric effects in the binding of these new compounds to HIV type 1 protease and a predictive CoMFA model was built with thirteen compounds as training set. Test analysis of other five compounds as test set demonstrated that the CoMFA model has strong predictive ability to this series of compounds. It will be very useful to further optimize the designed inhibitors.