Background: Enzymatic decolourization has been recently proposed as a promising and eco-friendly method for treatment of synthetic dye-contaminated wastewaters. However, the processes require large quantities of enzymes, attracting significant attention in developing efficient methods for mass production of multifunctional enzymes. Several methods such as response surface methodology (RSM) and orthogonal experiment have been applied to optimize the parameters in bioprocesses for enzyme production. Results: In the present study, a laccase-like enzyme, phenoxazinone synthase (PHS) originated from Streptomyces antibioticus was recombinantly expressed in Escherichia coli BL21 (DE3). The production of PHS in E. coli BL21 was optimized by response surface methodology based on Box-Behnken design. A full third-order polynomial model was generated by data analysis with Statistica 8.0 in which the optimal conditions for PHS production were calculated to be 1.525 mM CuSO4 and 16.096 hrs induction at temperature of 29.88ºC. The highest PHS production under optimal conditions was calculated to be 4098.51 U/l using the established model. Average PHS production obtained from actual production processes carried out under the calculated optimal conditions was 4052.00 U/l, very close to the value predicted by the model. Crude PHS was subsequently tested in Congo red decolourization which exhibited a low decolourization rate of 27% without mediator. Several mediators were found to improve PHS-catalyzed Congo red decolourization, with the highest rate of 73.89% obtained with 2,2’-azinobis-3-ethylbenzothiazoline-6-sulfonic acid (ABTS) as mediator under optimized conditions of 4000 U/l PHS activity, 10 μM ABTS, 100 μM Congo red, and 8 hrs reaction time. Conclusion: Our results indicated that PHS recombinantly produced in E. coli BL21 was a prospective enzyme for decolorizing reactive dye Congo red.