In this paper, we consider a two-step procedure for estimating the parameters in mass-balance based bioprocess models. In the first step, the values or combinations of the yield coefficients are calculated from mass-balance considerations without formulating any assumption with respect to the uncertain reaction rates. These values are then used, in the second step, into a general least-square identification problem which estimates the values of the other parameters and the missing initial conditions of the model. The estimation procedure is illustrated with an anaerobic digestion process considering experimental measurements from a fixed-bed digester during a process overload. This example demonstrates the applicability and the performances of the proposed approach. It also shows how model calibration can provide some hints to related biological questions.