With the rapid development of the computer capability and the continuous improvement of CFD (Computational Fluid Dynamics) technology, optimization design has gradually become a powerful and reliable means to solve the aerodynamic design problems. Among all kinds of optimization methods, gradient-based optimization algorithm can find the local optimal solution quickly, which is widely used in engineering problems. However, the computation of gradient is very expensive when an engineering problem involves large number of design variables. In this paper, an optimization method based on approximate gradient analysis is introduced, where the one-dimensional linear searches during the optimization process are evaluated with high-fidelity flow field analyses, while the gradients are evaluated with low-fidelity flow field analyses, leading to a significant saving in the computational cost of high-fidelity analysis. This method is demonstrated using a numerical example and RAE2822 airfoil optimization, and it is finally applied to the aerodynamic optimization of a wing-body configuration, which obtains good optimization results.