In order to investigate the basis for chiral separation in cellobiohydrolase 1 (CBH 1) and the closely related enzyme endoglucanase 1 (EG 1) from Trichoderma reesei, the wildtype proteins of CBH 1 and EG 1, as well as three catalytically deficient mutants of CBH 1 (E212Q, D214N and E217Q) were immobilised to silica and used as chiral stationary phases (CSPs) in HPLC. A large group of enantiomers could be completely resolved on the wildtype CBH 1-silica CSP while the corresponding EG 1-silica CSP only gave a partial separation of the same set of compounds. Of the CBH 1-mutant CSPs, only the D214N-CSP retained enantioselectivity whereas the selectivity was completely lost for the E212Q and E217Q-CSPs. The loss of enantioselectivity follows the same pattern as the loss of catalytic activity for the mutants which was determined from kinetic experiments using oligosaccharides as substrates. Mexiletine, a basic drug which could not be separated on the wildtype CBH 1-CSP, was successfully separated on one of the mutant phases. This demonstrates how protein engineering can be used to tailor new chiral selectors.