In this study, response surface methodology using Box-Behnken design was employed to study the effects of sucrose and nitrogen concentrations on pullulan production. Total of 15 experimental runs were carried out in a plastic composite support biofilm reactor. Three-dimensional response surface was generated to evaluate the effects of the factors and to obtain the optimum condition of each factor for maximum pullulan production. After 7-day fermentation with optimum condition, the pullulan production reached 60.7 g/L, which was 1.8 times higher than the result from original medium, and was the highest yield reported to date. The quality analysis demonstrated that the purity of produced pullulan was 95.2% and its viscosity was 2.5 centipoise (cP). Fourier Transform Infrared Spectroscopy (FTIR) also suggested that the produced exopolysaccharide was pullulan. Overall, this study demonstrated that the response surface methodology can be successfully applied to optimize medium composition for pullulan production in a biofilm reactor and maintained its high purity.