In order to clarify the differences of wind vibration response of base-isolated buildings calculated by three commonly used biaxial restoring force models of MSS, Casciati and Harvey and Gavin, three models were used to simulate the restoring force of lead-rubber bearing under horizontal uniaxial and biaxial displacement. Comparing the differences of tests or finite element results. The differences in base displacement, top displacement and top acceleration for a numerical example were analyzed using three models. The results show the trends that the restoring forces of three models simulate the lead-rubber bearing in uniaxial cyclic displacement, square and offset square displacement are basically the same. While simulating circular and offset circular displacements, the biaxial restoring force shape of MSS model is different from the finite element results, which cannot simulate the coupling behavior of the bearing accurately, and the error of Casciati model is slightly smaller than that of Harvey and Gavin model. The Casciati model and the Harvey and Gavin model are basically the same in calculating the wind vibration response. For the root-mean-square of the cross-wind response, the differences between the three models are not significant. For the root-mean-square of the along-wind base displacement, top displacement and top acceleration, the MSS model is slightly smaller, while for the peak factors of base displacement in along-wind and cross-wind, the MSS model is slightly larger. For the variation of the peak factor of top acceleration and the ratio of the maximum top acceleration of bidirection to unidirection with wind speed in along-wind and cross-wind, the MSS model differs from others. With the consideration of the simulation of biaxial coupling effect and the difference of wind vibration response, the Casciati model is suggested to consider the influence of biaxial restoring force model on wind vibration response of base-isolated buildings.