This article presents an efficient methodology for assessing the structural failure of micro-electromechanical systems (MEMS) used in safety and arming device (SAD) while taking into account the effect of fabrication errors. The proposed methodology uses statistical characteristics that are assessed using nonintrusive polynomial chaos expansion (PCE) to consider uncertainty. The effectiveness of the methodology is demonstrated by considering the example of the latching feature in MEMS SAD and comparing the results with experimental data obtained from a fabricated prototype. The proposed methodology accurately determines the failure probability and sensitive parameters of the sample, and provides insights into the impact of fabrication errors on the structural failure of the sample. It achieves the objective of reducing the burden of repetitive finite-element analysis (FEA) in uncertainty analysis and provides an efficient way to account for uncertainty in the study of MEMS devices.