The experimental investigation is conducted by collecting the vibration acceleration data of rolling bearings without any prior information on possibility distribution. The rolling bearing (type SKF6205) is used as an example to illustrate the application of maximum entropy grey bootstrap principle in analyzing the variation process. Monte Carlo method is applied to simulate the vibration acceleration data for different wear diameters. The upper and lower bound and estimated true value of reliability functions are achieved when the confidence level P=0.95. Then the variation probabilities can be calculated for different time series. The experimental investigation shows that the variation probabilities increase with the increase of wear diameters. Moreover, the variation probability presents a nonlinear increase trend, which can be divided into three stages: from the rapid increase stage to the slowly increase stage with trace fluctuations, and then to the rapid increase stage. The variation probability P n >0.8 when the wear diameter D=0.75mm and confidence level P=0.95, which shows that the variation is serious for the vibration performance of bearings. So the rolling bearings should be maintained or replaced to avoid the occurrence of serious accidents.