In this study, we present a multistage robust multiobjective optimization (MRMO) method that takes into consideration uncertainties stemming from fabrication errors. This efficient method has been employed to optimize an inertial setback feature (ISF) within the micro-electromechanical system (MEMS) safety and arming device (SAD). Sensitivity analysis and meta-model techniques are utilized to reduce the computational dimension and cost of optimizing the objectives. The accuracy of the approximation between the global meta-model (GMM) in the method and finite element analysis (FEA) results is guaranteed through appropriate calculation samples. Two optimization algorithms, the fuzzy multiobjective particle swarm optimization (f-MOPSO) algorithm and the adaptive accelerated gravitational search algorithm (AAGSA), are performed to global optimization and local search for the fitness values sequentially to enhance the optimization performance. Robust optimal solutions are shortlisted and selected from among candidate solutions exhibiting superior performance. The effectiveness of the method is demonstrated through a comparison of the results obtained from FEA and the GMM with experimental data collected from some fabricated ISF prototypes. This method achieves the objective of reducing the burden of repetitive FEA calculations and better performance enhancement in robustness optimization.