This work deals with the methodology put in place to fit and validate the parameters of a biofiltration model (BAF) in tertiary nitrification treatment and dynamic conditions. For an average loading rate of 0.65 kg NH4-N/m3 media/d, different time loading rates are applied inside a filtration-backwash run using a semi-industrial pilot. Comparisons between predicted and observed values on the NH4-N, NO3-N and TSS in treated water and the total head loss DP are carried out firstly using default values of BAF parameters. Model predictions overestimate values measured but trends are well reproduced. A sensitivity analysis is carried out and the hierarchy of BAF parameters has been set up classifying them into strong and low influence on the effluent concentrations. Among parameters revealing the strongest influence are those of the filtration module and the mean density of biofilm for the TSS effluent and the total DP, the specific autotrophic growth rate, the maximum biofilm thickness and the reduction coefficient of diffusivity in the biofilm for the NH4-N, NO3-N effluent. Finally, this classification leads to setting a calibration procedure, thanks to specific experimental tests directly measuring some BAF parameters.