This paper deals with the comparison between two control strategies for the optimal operation of Activated Sludge Processes (ASPs). The first approach describes an industrial control strategy composed of two decentralized Proportional Integral controllers (PIs) while the second approach is based on a state space model of the system together with the Disturbance Modeling Principle (Johnson, 1976). It is shown that if the second approach needs additional sensors than the first one, when implemented properly, it allows the user to estimate unknown inputs that can be useful for diagnosis purposes. Simulation results are provided before some conclusions are drawn.