This paper proposes a model-based control strategy that can predict the influent and effluent as well as control the effluent water quality in the A(2)/O (Anaerobic/Anoxic/Oxic) process for 1 day in advance. In the model-based control strategy, ANN (Artificial neural network) and modified ASM3+Bio-P model were used to predict the influent and effluent for 1 day in advance, respectively. When the predicted effluent NH4-N concentration was higher than the target value, the optimal setpoint schedules of the DO (Dissolved oxygen) could be deduced using a scenario simulation. The scenario simulation was carried out to obtain DO setpoint schedules to reach the effluent NH4-N concentration under the target value. The deduced optimal setpoint schedules were used to control the operation of the process for the next day. The results of model prediction showed that the behavior of the influent and effluent could be predicted successfully. The proposed model-based control strategy was tested in a pilot-scale A(2)/O process for 2 weeks, which confirmed that the effluent NH4-N concentration could be maintained steadily lower than the target value of 5 mg/L. The air flow rate during control period was increased by 9% of that during without control period. On the other hand, their corresponding average effluent NH4-N concentrations were 4.42 and 13.89 mg/L, respectively, which highlight the significant effect of this control strategy with only a slight increase in air flow rate. These results show that the developed model-based control strategy can be used successfully in the A(2)/O process and possibly other processes. (C) 2012 Elsevier B.V. All rights reserved.