An effective interatomic potential is crucial for molecular dynamics simulations. In this paper, a more effective optimization method for potential parameters is proposed. Sensitivity analysis is used to find the key parameters affecting the structure and properties most in all the potential parameters. The secondary parameters are ignored, and the dimension of the optimization is reduced. The genetic algorithm is adopted to optimize the key parameters. The results show that the structure and physical properties of BaTiO3 and SrTiO3 simulated by the optimized potential are more close to experimental data.