In order to achieve successful long-term embedding of Micro-Electro-Mechanical Systems (MEMS) sensors, issues pertaining to their reliability and failure modes need to be investigated. Understanding critical failure modes and their influence on the sensor output is key to developing micro-sensors and switches for long-term application without re-calibration or periodic device maintenance. In this effort, a hierarchical and multi-level model is developed for simulating the response of a MEMS device under induced mechanical failure conditions. The example MEMS device, an accelerometer is modeled at the dynamic, electro-mechanical interaction, and device levels. Mechanical dynamics level simulates the deformation and stress response of the mechanical stage to applied loads. Electro-mechanical interaction model is used to simulate the electrostatic-mechanical interaction of an accelerometer predicting the capacitance changes and electrostatic forces due to deformation. Device level emulates the signal analysis and calibration stage, thus enabling the simulation of the electrical output when the device is subjected to mechanical force or acceleration inputs. The modeling of the MEMS device at each level is carried out using the Finite Element Method. Potential failure modes, such as internal delamination of the mass and spring elements and micro-cracking, are then emulated. The effect of each failure mode on the sensor output is studied for an identifiable signature in the sensor output signal. Selected results showing the effects of typical failure modes are shown.