The complex thermal coupling between complementary metal-oxide semiconductor (CMOS) components and other modules of an in-orbit assembled telescope and significant changes in external heat flow make it difficult to determine the optimal parameters for the thermal design of CMOS components. In this study, a hybrid sensitivity analysis method was developed based on the mean impact value(MIV) algorithm and the traditional regression analysis (Pearson and Spearman) algorithms for comparison and validation. In addition, a sensitivity analysis of the thermal design parameters of the CMOS component was conducted, the optimal parameters were obtained, and thermal design optimization was achieved. Compared with the traditional traversal method for selecting thermal design parameters, the number of thermal design parameters decreases from the original 10 to two important and five minor parameters. This makes the selection of parameters purposeful and improves the thermal design efficiency. The simulation results show that the temperature fluctuation of 1.6℃–25.4 ℃ during the CMOS track cycle and the thermal design satisfy the operating temperature requirements, verifying the feasibility of the thermal design based on the hybrid sensitivity analysis method.