Capsid protein enterovirus 71 (EV71) is one of the major viruses that cause the severe encephalitis and thus result in a high mortality in children less than 5 years of age.In an effort to discover new potent inhibitors against EV71,a novel three-dimensional pharmacophore model was developed on 24 inhibitors with different molecular structures and bioactivities.The best hypothesis (Hypo1) has a high predictive power and consists of four features,namely,one hydrophobic point (HY) and three hydrogen-bond acceptors (HA).Two key features of the best Hypo1,HY1 and HA3 match well with an important narrow hydrophobic canyon and with the surface of LYS274 in the target EV71 active site,respectively.The more versatile feature,HA1,is firstly found to be very influential on these compounds’ bioactivities,which may interact with the other side of the active site in the EV71 receptor.The application of the model is successful in predicting the activities of 30 known EV71 inhibitors with a correlation coefficient of 0.831.Furthermore,Hypo1 demonstrates a superior screening capability for retrieving inhibitors from the database with a high enrichment factor of 70.This study provides some important clues in search for more potent inhibitors against EV71 infection. Capsid protein enterovirus 71 (EV71) is one of the major viruses that cause the severe encephalitis and thus result in a high mortality in children less than 5 years of age. In an effort to discover new potent inhibitors against EV71, a novel three-dimensional pharmacophore model was developed on 24 inhibitors with different molecular structures and bioactivities. The best hypothesis (Hypo 1) has a high predictive power and consists of four features, namely, one hydrophobic point (HY) and three hydrogen-bond acceptors key features of the best Hypo1, HY1 and HA3 match well with an important narrow hydrophobic canyon and with the surface of LYS274 in the target EV71 active site, respectively. more versatile feature, HA1, is the first found to be very influential on these compounds ’bioactivities, which may interact with the other side of the active site in the EV71 receptor. The application of the model is successful in predicting the activities of 30 known EV71 inhibitors with a correlati on coefficient of 0.831.Furthermore, Hypo 1 demonstrates a superior screening capability for retrieving inhibitors from the database with a high enrichment factor of 70. This study provides some important clues in search for more potent inhibitors against EV71 infection.