Aerodynamic shape optimization involving a complex geometric model or problem may have tens or hundreds of design variables, necessitating multiple accurate but time-consuming computational fluid dynamics simulations to produce optimal designs, which greatly affects the efficiency of optimization and. To address this challenge, this article proposes an efficient optimization method based on design-variables-screening. Within the framework of the method, a complicated input–output relationship is broken down into quantitative effects. The influence of design variables on the objective function is calculated by the Kriging regression model and functional analysis of variance. In the meantime, a screening strategy is proposed to facilitate the selection of design variables for optimization. The less important design variables in the problems of interest are fixed so that the dimensionality of the problems is reduced to save computational cost. Experimental results on the National Advisory Committee for Aeronautics airfoil (NACA0012) demonstrate that the simplified model with the screening strategy achieves nearly the same reduction in drag coefficient as the conventional method that optimizes all design variables. Moreover, it significantly enhances the efficiency of optimization and contributes to the enhancement of flow stability.