In this paper, according to the degradation data characteristics of MEMS (Micro-Electro-Mechanical Systems) accelerometers of small samples, the uncertainty theory is introduced into, and the reliability model of the product is established by deduction according to the three basic scientific principles of reliability analysis. In this paper, the parameter estimation and model validation problems are solved under the framework of uncertainty theory by using the great likelihood estimation and hypothesis testing method under the algorithm of uncertainty theory. It is analyzed that the MEMS accelerometer reliability model based on uncertainty theory is a new method for reliability assessment applicable to dealing with uncertainty problems in the absence of large samples of degraded data.