A model updating approach based on reduced models is developed. This approach is based on the structural sensitivities of the system mass and stiffness matrices, and the structural parameters are adjusted. The analytical mass and stiffness matrices of the reduced models can be adjusted simultaneously. After several iterations, the modal data of the updated reduced model reproduce those of the test model. The scheme to transform generalized coordinates into physical coordinates is also described. In the case of reduced models, error is introduced by the process of model reduction itself. Thus, the analytical mass error effect may not be apparent and needs further exploration.