Optically pure methyl (R)-o-chloromandelate and (R)-acetyl-o-mandelic acid are key intermediates for the synthesis of (S)-clopidogrel, which could be prepared with 100 % theoretical yield by sequential hydrolysis and racemization. At the moment, efficient sequential hydrolysis and racemization are hindered by the low catalytic activity of mandelate racemase (MR) toward (S)-o-chloromandelic acid ((S)-2-CMA). In the present work, we proposed to improve the catalytic performance of MR toward (S)-2-CMA by directed evolution and developed an enantioselective oxidation system for high-throughput screening (HTS) of MR libraries. Based on this HTS method, a triple mutant V22I/V29I/Y54F (MRDE1) with 3.5-fold greater relative activity as compared to the native MR was obtained. Kinetic analysis indicated that the enhanced catalytic efficiency mainly arose from the elevated k (cat). Further insight into the source of improved catalytic activity was gained by molecular simulations, finding that substrate binding and product release were possibly made easier by decreased steric bulk and increased hydrophobicity of substrate binding sites. In addition, the substrate (S)-2-CMA in the enzyme-substrate complex of MRDE1 seemed to have a lower binding free energy comparing with the complex of wild-type MR. The HTS method developed in this work and the successful directed evolution of MR based on this method provide an example for racemase engineering and may inspire directed evolution of other racemases toward enhanced catalytic performance on non-natural substrates.