In this paper, regardless of the cooling system of the mold, the optimization design for the material distribution of heterogeneous injection mold is investigated, which is made of two base materials with different thermo-mechanical properties. A mathematical model is constructed in which both a function related to casting average temperature and the temperature gradients throughout all the cavities during the solidification process are taken as the objective function while the von Mises stress constraints of the mold are considered for reducing the stress concentration in the mold. The Finite element method (FEM) and the finite difference method (FDM) are utilized to analyze the transient heat conduction of the mold and casting. The constrained optimal design problem is solved using real coded genetic algorithm. A smooth bicubic B-spline interpolation is proposed for representing the continuous variation of material distribution. The effectiveness of the optimization results is verified by a numerical example.