Abstract: A FE Model updating approach based on Gaussian white noise mutation particle swarm optimization (GMPSO) is proposed to improve the efficiency of the model updating method and keep its accuracy. The particle swarm optimization (PSO) and the improved method GMPSO were introduced and the global optimization searching capacity was compared between the two methods. The FE model updating method based on GMPSO was then proposed and the one-to-one match between the parameters in the method and the real physical variables was discussed together with the method process. The new approach was applied to the model updating of a damaged simply supported beam (DSSB) with high-dimension (with ten variables) and the updating results was compared with that of model updating method based on Genetic Algorithm (GA). Meanwhile, the method is also applied to an in-serviced bridge structure (with thirteen variables) to validate the feasibility of the proposed method. The results show that:the global optimization searching capacity of the improved GMPSO is significantly higher than that of PSO; the relationships between the parameters and physical variables are clear in the model updating method based on GMPSO and this is suitable for modularized program operation and facilitate the application of commercial software. The updating process of the DSSB with high-dimension using the proposed methodology reveals accurate results and its efficiency is significantly improved, compared with the process of model updating based on GA. The updating of the in-serviced bridge using the proposed methodology narrows the differences between the main girder vibration frequencies of model and those of test. The feasibility of the application of the proposed model updating methodology to the updating issue of high-dimensional and complex engineering structures is verified.