Abstract: During the operation of solid oxide fuel cell （SOFC）, complex physicochemical phenomenas such as convection, diffusion, surface reactions, and charge transfer reactions will occur. Coupling the reaction kinetics model with electrode simulation model can predict the performance of SOFC. Compared to the Butler-Volmer equation, the multi-step elementary reaction model can better describe the actual electrode kinetics. However, parameters of multi-step elementary reaction model usually contain significant uncertainties, which affects the accuracy of model predictions. To reduce model prediction uncertainty, an anode model for SOFC using humidified hydrogen gas （H₂/H₂O） as fuel is eatablished in this study, and polarization curve of the anode is calculated. Sensitivity analysis is conducted on the kinetic and thermodynamic parameters, with 11 sensitive parameters being identified. Forward and reverse uncertainty analysis are performed on the anode model separately, and the model prediction performance is optimized based on the results of uncertainty analysis. The results show that the optimized anode model reduces the prediction errors of the polarization curves at temperatures of 1 023.15 K and 1 123.15 K from original 33.12% and 34.51% to 8.61% and 15.47% respectively, the model prediction accuracy is improved.