The sulfur autotrophic reduction (SAR) process is promising in co-reduction of perchlorate and nitrate from aqueous solution. To further understand the reaction process, we developed a sulfur autotrophic fluidized bed reactor where the proceeding extent of sulfur (S) disproportionation was predicted by Response surface methodology (RSM) for the first time. Three fundamental reaction parameters including the hydraulic retention time (HRT), co-existing nitrate concentration (\({C}_{\mathrm{inf }{NO}_{3}^{-}-N}\)) and recirculation ratio (R) were considered for reactor optimization. The results demonstrated that S disproportionation was promoted by long HRT and high R, whereas was inhibited by high \({C}_{\mathrm{inf }{NO}_{3}^{-}-N}\). Also, the optimal HRT, \({C}_{\mathrm{inf }{NO}_{3}^{-}-N}\) and R were 0.50 h, 10.00 mg/L and 14, respectively, the bioreactor can achieve high reduction efficiency of perchlorate and nitrate (> 98.45%), and generate less sulfate (236.07 mg/L). High-throughput sequencing showed that Chlorobaculum was related to S disproportionation, and Sulfurovum was associated with nitrate/perchlorate reducing. All results indicate that the sulfur autotrophic fluidized bed reactor is a promising candidate for the treatment of perchlorate and nitrate wastewater in future practical applications.