As for a novel micro-manipulation stage, a reliability-based robust optimization design method was proposed based on 6σ robust design and response surface method considering the random fluctuations of its structural parameters. The design variables were determined by using screening experiment. The response surface models, which reflect performance indexes of stage, were built by using the central composite design. The first-order second moment method was used to obtain the robustness and reliability indexes of the stage based on the model, such as the probability distribution of the displacement amplification factor, natural frequency and reliability sensitivity, and reliability. Based on the six sigma robust design, the reliability-based robust optimization model was proposed as the robustness of the displacement magnification factor, the natural frequency and the reliability sensitivity for the objective, and the reliability of the stage for constraint. The sequential quadratic programming method is adopted to solve the optimization problem. Comparing the results of the robust optimization with deterministic one, the results showed that the variance of the displacement magnification, the natural frequency and reliability sensitivity of the stage were reduced by 40.9%, 49.4% and 33.6% respectively, and the reliability increased from 61.36 % to 99.78 %. Thus, it meets the design requirements of reliability-based robust optimization, and it illustrated that the proposed design ideas and theoretical model are effective.