Abstract: A high-efficiency Pseudomonas sp. was separated from oil-contaminated soil and its dehydrogenase activities toward anthracene and pyrene were investigated. Based on the amount of dehydrogenase (DA), the Box-Behnken model and response surface methodology (RSM) were adopted to screen and optimize the fermentation conditions, i.e., temperature, salinity and weight percentage of quality gradient. The results showed that the P values of these models were all less than 0.05, indicating the models were significant and fitted well with experimental data. For anthracene, the optimal dehydrogenase conditions were:3.89% salinity, 35.73℃, 5% quality gradient. Under these conditions, the observed DA was (140.353±6.430)μg, which was close to the predicted value of 141.466 μg. Hence, the optimized treatment conditions were reliable. All factors could significantly influence the DA, and the order of importance was temperature > quality gradient > salinity. However, the optimal dehydrogenase conditions for pyrene were:0.73% salinity, 34.78℃, 7% quality gradient. Under these conditions, the observed DA was (84.032±0.063)μg, which was close to the predicted value of 86.304 μg. Hence, the optimized treatment conditions were also reliable. All factors could significantly affect the DA, and the order of importance was quality gradient > salinity > temperature. It can be seen that temperature is the main factor that determines the degradation of anthracene, while quality gradient is the main factor that controls the degradation of pyrene. Our results showed that these models were reliable to predict the DA of the Pseudomonas sp. and were suitable for industrial application.