Superconducting magnetic energy storage (SMES) systems with different superconducting materials are attracting great attentions and funding from the governments around the world because they are promising large-scale energy storage devices for future smart grid. Due to the high cost of SMES, its manufacturing quality and operation reliability have to be investigated in the design optimization stage. This paper presents a robust design optimization framework to solve this issue based on a benchmark problem, TEAM problem 22. The proposed robust design optimization is based on a technique called design for Six-Sigma. Meanwhile, a modified multilevel optimization strategy is employed to reduce the computation cost of finite element analysis due to high-dimensional design space and Monte Carlo analysis. As shown, the reliability and manufacturing quality of the investigated SMES after robust optimization have been increased greatly. Index Terms-Manufacturing quality, multilevel optimization, robust optimization, superconducting magnetic energy storage.