Abstract The properties inhomogeneity of steel plates will result in uncertainty of forming quality of large diameter UOE oil pipes and increase the risk of pipeline failure. In this paper, a robust design method for UOE forming process based on support vector machine and sequential response surface modeling and considering the variation of steel plates′ properties is proposed. A hundred of mechanical experiments are first carried out and the variation of X80 steel's properties is statistically evaluated. The varied properties are assigned to partitioned steel segments and taken as the noise factor in process design. The ovality of the UOE pipe is employed as the optimization objective, and the forming quality indicators including convex amount, mean outside diameter, yield strengths and Ys/Uts7, O-forming gap, the width of the U-shaped plate, inclined angle of straight arm of the U-shaped plate, etc., are taken as the constraints. Based on the semi-analytical computation of C-, U- and O-forming processes and finite element simulation of expanding process, a sequential response surface model is established by using the support vector machine method. Finally, a Monte Carlo sampling is performed to demonstrate the effectiveness of the proposed method. Compared to conventional optimization method, the robust optimization performs obviously better in reducing the ovality and increasing the robustness of UOE forming process.