Abstract: Conventional tests and theoretic calculations are unable to analyze the effects of various random errors on relevant dynamic characteristics. A virtual test was therefore proposed for the reliability of rolling bearings. Based on dynamical contact mechanics and FEM of explicit dynamics, a 3-D parameterized FEM model was developed for ball bearing with deep filling slots. Then, a numerical simulation was done using the software LS-DYNA for the operating process of ball bearings to give the contact stress/strain variation between a bearing's inner race, outer race, balls and ball cage, as well as the pressure distribution in contact process. And the computation of pseudo-random numbers was programmed by multiplying the congruent number with the APDL language used. Taking account of the effects of original errors resulting from manufacture and the different working conditions such as rotating speed and load on the dynamic property of a bearing, the K-S inspection was done to determine the distribution of the data resulting from the virtual test and, consequently, the reliability was calculated with respect to the fatigue failure of bearings. Furthermore, the simulative computation was done 1000 times for the bearings with the Monte Carlo method to give the sensibilities of reliability of all parameters, thus providing theoretically the reference for both the calculation of bearing's fatigue strength and dynamically optimum design.