Abstract: An elasto-plastic finite element method is developed to predict the thermal residual stress of thermal spraying coatings with functionally graded material (FGM) layer. The optimized distribution form and parameter p about the volume fractions of various constituents in the FGM are obtained by the first order optimization method in the AL₂O₃ -Ni model system. The effects of geometry and material behavior on the optimization results are investigated numerically, including the temperature sensitivity of FGM and substrate material. It is found that the optimization of the constituent contents in FGM reduces the magnitude of residual stresses to a large degree. And the maximum residual stresses shun the weakest part of the coating structure by the optimization designing. When the length of specimen, the thickness of FGM layer and the thermal expansion coefficient of the substrate increase, while the distribution parameter p decreases. The results presented in this paper are useful for the design of thermal spraying coatings.