Ursodeoxycholic acid (UDCA), a pharmaceutical ingredient widely used in clinics, can be prepared from chenodeoxycholic acid (CDCA) by the epimerization of the 7α-OH group. In this study, a nicotinamide adenine dinucleotide (NAD + ) regeneration system was developed by using flavin oxidoreductase (FR) and flavin mononucleotide (FMN). Only catalytic amount of FMN is required for the effective NAD + recycling. FR/FMN system was then applied in the oxidation of CDCA to 7-ketolithocholic acid (7-keto-LCA) by NAD + -dependent 7α-hydroxysteroid dehydrogenase (Bs-7α-HSDH) from Brevundimonas sp., which showed extremely high enzyme activity toward CDCA (k cat /K m = 8050 s À 1 · mM À 1 ). When Escherichia coli whole cells coexpressing Bs-7α-HSDH and FR genes were used as biocatalyst, CDCA (50 mM) was completely converted to 7-keto-LCA with the turnover number of FMN being 227 and 58.8 g · L À 1 · d À 1 space-time yield of 7-keto-LCA. For the reduction of 7-keto-LCA, nicotinamide adenine dinucleotide phosphate (NADPH)-dependent 7-β-hydroxysteroid dehydrogenase (Cm-7β-HSDH) from Clostridium sp. Marseille was employed with alcohol dehydrogenase from Thermoanaerobacter brockii (TbADH) and iso-propanol as co-factor regeneration system. When E. coli whole cells coexpressing Cm-7β-HSDH and TbADH genes were used as biocatalyst, 40 mM 7-keto-LCA was reduced to UDCA with 26.8 g · L À 1 · d À 1 space-time yield. The oxidation and reduction were then carried in a one-pot concurrent mode, 12.5 mM CDCA was completely converted to UDCA. The epimerization of CDCA to UDCA proceeded to completion at the substrate concentration of 30 mM in the one-pot sequential process. Therefore, the complete conversion of CDCA to UDCA in one-pot has been realized by employing 7α-HSDH and 7β-HSDH of different co-factor specificities with independent co-factor recycling systems. The cholic acids, especially UDCA, exert inhibitive effect on the activities of these enzymes, preventing the complete epimerization of 7α-OH at higher substrate loading. This inhibition issue should be solvable by engineering the involved enzymes, that is currently pursued in our laboratory.