This dissertation proposes a novel accumulative test method for MEMS-based, commercial off-the-shelves inertial measure units (IMUs) from the automotive field. The need for clarification and augmentation of the corpus of knowledge on MEMS reliability for space applications is addressed. Faster and less tedious test procedures, better understanding of the failure modes as well as the possible prediction models and numerical simulations are at the center of a pragmatic vision that shall keep the industrial end-user and applications in sight. The IMU is used in several sequential accumulative test campaigns under various environments. The device comprises notably of a X,Y-axis accelerometer. The structure of this research is elaborated following an increasing level of complexity, culminating at a seven-step stress test, alternating between thermal shocks and vibration testing, followed by thorough failure modes and effects analysis of the failed or degraded devices. Chapter 1 describes the world of MEMS. Highlights of uses in space technologies are presented, as well as the state-of-the art of reliability testing and corresponding failure mechanisms. Remaining open questions about accumula-tive testing are covered. Chapter 2 elaborates on the stakeholders implied in this research. It is important to keep in mind the highly applic-ative and industrially oriented vocation of reliability testing. Choice of the IMU for the various test campaigns is done. A thorough destructive physical analysis has been performed to gain knowledge of the devices' construction. Chapter 3 details the experimental method. Functional tests were performed before, in-between, and after a test step on the devices alone (no peripheral circuitry). Testing methodology description follows with the development of a mono-parameter, two-step stress test procedure. Such procedure fails to properly assess the reliability of the robust MEMS components, even with harsher environments than commonly used standards (MIL-STD-883). Accu-mulative testing was therefore performed. Chapter 4 reports the experimental results from the different test campaigns: (i) temperature cycling followed by vibration, (ii) thermal shocks followed by vibration, (iii) vibration and thermal shocks alternated in a seven-step sequence. The first campaign yielded to the failure of the devices due to die attach weakening, while the two oth-ers showed intense degradation without failing. X-ray tomography was used to identify failure modes, as well as high-resolution X-ray diffraction for residual stress identification in the silicon dies. Other complementary tests were performed (die shear, wire bonds pull). In Chapter 5, a finite element model of the MEMS device is designed. A thermal-only model allows to verify that devices are at thermal equilibrium during testing. Then, by relying on a viscoelastic, stress and strain data can be obtained over a thermal cycle for the die attach. The found values suggest that accumulation of plastic damages are very likely to occur due to the build-up of thermal stress in the adhesive: a result corroborated by the experi-mental observations. By proposing a novel approach to robust MEMS reliability approach and diving deep into the failure modes and effects of accumulative testing, this research aims at contributing to the elaboration of the next framework that will help the space community to foster a wider use of COTS devices in space applications.