Abstract: The seismic demand of bridge structures is affected by the randomness of the ground motion as well as the random parameters of the structure. In order to analyze the importance level of each random parameter, the importance analysis method is proposed to rank the random parameters involved in bridge structures. Based on the nonlinear time-history analysis of structural random samples, the simply supported and continuous beam bridge are taken as examples to show the calculation of variance-based importance measures and moment-independent importance measures of each random parameter by Monte Carlo sampling and kernel density estimation respectively. The results show that the importance rankings of random parameters are not the same for the seismic demand of piers, abutments and bearings in medium or small span continuous girder bridges. However, the parameters such as bearing shear modulus, mass of upper structure and damping ratio are ranked high for the seismic demand of all components. Compared with local sensitivity analysis methods such as Tornado graphic method, the importance analysis method can consider the influence of other random parameters in studying the importance level of one variable, and its application is more reasonable.