In the work, molecular docking method was applied to extensively predict the enantioselectivity of lipases and esterases. A ligand library consisted of 69 chiral substrates was docked to four lipases and two esterases to set up the prediction model. During the docking process, necessary modification was carried out on van de Waals and hydrogen bond parameters of enzyme/substrate pair so that the ligands were able to adopt productive geometry in the enzymes. The docking results correctly indicated the enantiopreference for 91% (63/69) of docking pairs and the docking energy difference between Substrate enantiomers (Delta Delta Gd-docking) was significantly (correlation coefficient=0.72, P<0.05) correlated with the activation free energy difference (Delta Delta G not equal) that was quantitatively correlated with enantioselectivity of the enzymes. The prediction method was further validated by docking with another 12 enzyme/substrate pairs. Moreover, the prediction error was susceptible to the size of groups bonded to substrate's chiral center and expected Delta Delta G not equal Values but was not related to the substrate type and reaction medium. The possible reasons of observed error were discussed. It is demonstrated that the docking method has great application potential in high performance prediction of enzyme enantioselectivity. Biotechnol. Bioeng. 2010;105: 687-696. (C) 2009 Wiley periodicals, Inc.