In order to improve the stamping formability of thin shell parts for 3C  magnesium alloy products, the magnesium plate temperature and punch temperature were introduced into the optimization process, and the stamping process was optimized by the combination of orthogonal experiment and multi-objective particle swarm optimization (MPSO) algorithm. Then, on the basis of orthogonal experiments, a quadratic polynomial response surface model with cracking and wrinkling as the objective functions was constructed, and the Pareto solution set was obtained by multi-objective particle swarm optimization algorithm. Furthermore, the optimal particle Pm of multi-objective particle swarm was obtained by the minimum distance selection method, namely, the die fillet radius Rc is 6.90 mm, the magnesium plate temperature Td is 276.33 ℃, the punch temperature Tp is 64.69 ℃, and the blank holder force F is 22004.14 N. The simulation and experimental results show that the cracking objective function Robj for the optimal particle Pm of multi-objective particle swarm is 22.96% lower than that for the optimal sample in orthogonal experiment, and the wrinking objective function Wobj is reduced by 16.43%. It shows that this method can effectively solve the multi-objective optimization problem of differential temperature stamping for magnesium alloy thin shell parts.