An industrial-scale dead-end ultrafiltration system was optimized using statistically designed experiments. Given a certain level of pollutant, a two-level full factorial design and a central composite design were used to optimize the filtrate production of a single 8-inch industrial ultrafiltration membrane while manipulating the levels of four factors: feed pressure, backwash pressure, forward filtration time, and backwash time. Analysis of variance and residual analysis were used to validate and check the adequacy of the developed regression models. The optimal levels were later validated experimentally. The predicted filtrate production was in reasonable agreement with the experimental data.