Automotive passive safety optimization design (APSOD) is a multi-dimensional, complex-constraints nonlinear multidisciplinary optimization problem. Simulation model based optimization (SMO) and approximation model based optimization (AMO) are two classical optimization methods for APSOD. However, the randomness of SMO limited its use in APSOD, which could generate the great number of near-duplicate or imperfect design points. On the other hand, AMO has relative low accuracy, because of the influence of the disturbed data in training dataset. Therefore, this study intends to present an new improved multidisciplinary optimization search strategy in APSOD. Firstly, the heuristic optimization strategy in SMO is employed, and a clustering analysis is added to identify and delete the near-duplicate or imperfect design points generated in the heuristic optimization process. Then, the remaining ones are used to build an approximation model to perform AMO, and the quadratic optimized data, in turn, are further used to perform SMO until the stopping criterion is satisfied. This proposed method, which integrates with SMO and AMO process, has been applied to the car body design, and the superiority of this integrated multidisciplinary optimization search strategy has been validated by comparing with classical SMO method such as NSGA-II.