The cooling system of a plastic injection mold plays a crucial role in determining both the productivity of the injection molding process and part quality. The cooling process is an important phase of injection molding, which could account for more than three-fourths of the total molding cycle time. Thus, an efficient cooling will significantly improve productivity by reducing the cooling time, and uniformity and balanced cooling can ensure part quality by preventing sink marks, different shrinkage, internal thermal residual stress and warpage. In this paper, the geometric parameters (i. e., radius and location of each cooling channel) and process parameters (i. e., inlet coolant temperature, volumetric flow rate of each cooling channel, packing time, and cooling time, etc.) are considered systematically in the cooling system design of injection molding. A two-level decomposition method for the cooling system optimization is proposed. The optimization problem is divided into two subproblems (cooling channel design and process design) on the basis of the different effects on refrigerant behaviour of different kinds of design variables. By combining the Kriging model and CAE technology, a mold cooling system for a 15-inch display was optimized. The results show that the method can efficiently improve the temperature distribution and the part quality.