A validated 3D pharmacophore model was generated for a series of ACE inhibitory peptides, which consisted of five features (two hydrophobic functions, two hydrogen bond acceptors, and a negative ionizable function). The built model was able to correctly predict the activity of known ACE inhibitors. The model was then used as query to search 3D databases of peptides. Three novel peptides (I, II and III) were synthesized and biologically evaluated in vitro. It appears that the in vitro activity of peptides I, II and III was consistent with their molecular modeling results. Our results provided confidence for the utility of the pharmacophore model to retrieve novel ACE inhibitory peptides with desired biological activity by virtual screening.