This article first proposes a reduction strategy of the activated sludge process model with alternated aeration. Initiated with the standard activated sludge model (ASMI), the reduction is based on some biochemical considerations followed by linear approximations of nonlinear terms. Two submodels are then obtained, one for the aerobic phase and one for the anoxic phase, using four state variables related to the organic substrate concentration, the ammonium and nitrate-nitrite nitrogen, and the oxygen concentration. Then, a two-step robust estimation strategy is used to estimate both the unmeasured state variables and the unknown inflow ammonium nitrogen concentration. Parameter uncertainty is considered in the dynamics and input matrices of the system. (C) 2009 American Institute of Chemical Engineers Biotechnol. Prog., 25: 654-666, 2009