Abstract: The issues related to using the response surface method for finite element model updating such as experiment design, parameter selection, response surface regression and optimization are summarized firstly. Then the finite element model of a rigid frame-continuous girders bridge, Jingyuan Yellow River Bridge, is updated based on the in-situ dynamic test results. The elastic moduli of the box girder and the pier, transverse and vertical stiffness values of the bearings are selected to be the parameters for updating. The central composite design method is chosen to conduct the experimental design and calculate the sample values. Two response surface models with and without considering the influence of cross-terms are constructed respectively and used to be updated instead of updating the initial finite element model. It can be concluded that the cross-terms have less influence on the updating parameters and little influence on the modal frequencies. The analytical frequencies from the updated models are in good agreement with the identified frequencies based on dynamic test. The updated finite element model can accurately represent the dynamic characteristics of the bridge. Therefore it can serve as the benchmark model for further damage identification and condition assessment of the bridge.