MEMS technology has enabled substantial expansion of the inertial sensor market by decreasing power consumption, size, and cost of inertial sensors. Furthermore, with significant performance improvements in recent years, MEMS inertial sensors are used in many emerging applications, including autonomous systems, position-sensing and navigation, space, and defense. However, maintaining high performance of MEMS inertial sensors remains challenging as it is not perfectly resistant to changes in surrounding temperature or other environmental effects. In this study, a single-isolation stage oven-controlled inertial sensor (SOCIS)---an oven-controlled inertial sensor using one isolation stage---is implemented to mitigate environmental effects. The isolation stage is suspended beam structure and made of low thermal conductivity material to isolate mechanical and thermal effects. A commercial IMU, Invensense MPU-6050, is attached to the isolation stage and vacuum-sealed in the LCC package. Furthermore, to control IMU temperature accurately, different temperature sensing methods are studied, among which the method representing the ‘’actual’’ IMU temperature most accurately is selected. Experiment results during thermal-cycle show temperature effects on gyroscopes and accelerometers are improved by 108 and 13.1 times, respectively, for the best SCOIS, while average improvements of eight SOCISs are 11 and 1.1 times, respectively. Furthermore, experiment results during strain-induced stress tests show the stress effect on gyroscopes is improved by 36 times. Finally, simulation results suggest that the SOCIS would filter mechanical vibrations at 3.1kHz or higher. Although SOCIS mitigated environmental effects significantly, the system is limited by two fundamental limitations: temperature-induced stress, which is caused by temperature gradient and its change on the glass isolation stage, and limited design and material selections for the isolation stage. Therefore, a new structure, a double-isolation stage oven-controlled inertial sensor (DOCIS), is proposed to address these limitations. In DOCIS, the temperature gradient is minimized by replacing glass with silicon. Since such isolation stage cannot provide a good thermal isolation capability, a second stage is added between the silicon stage and LCC package to provide thermal isolation. The second stage is made of cross-linked polyimide aerogel for extremely low thermal conductivity of 0.03W/mK, and higher mechanical durability than that of typical silica aerogel by 100 times. Furthermore, to increase the thermal resistance of DOCIS further, the contact area between the silicon and aerogel isolation stages is minimized by forming small bonding pads at the bottom surface of the silicon isolation stage, and electrical connections are established using 0.5-mil platinum bonding wires. Simulation and experiment results show that thermal resistance at 1mTorr is 1,700K/W and the maximum temperature-induced stress on inertial sensors would be reduced by 62 times. Finally, DOICS packaging technology is applied to a fused-silica birdbath shell resonator to mitigate the temperature effect on the resonator and to survive a 5,000g impact. A 9x9 array of square bonding pads is formed at the bottom surface, and no suspended structure is used in the silicon isolation stage. Simulation results show DOCIS will survive a shock impact higher than 5,000g when the width of the square bonding pad is larger than 800um. Finally, total thermal resistance would be 817K/W, when the aerogel stage is 250um-thick, 22 electrical connections are established using 0.5-mil platinum wires, and the pressure in the package is lower than 1mTorr. This can be increased to 1180K/W and 1520K/W, if the thickness of the aerogel stage is increased to 375um and 500um, respectively.