For the rolling bearing with expensive test cost or of inconvenient test, in order to efficiently and accurately analyze its contact fatigue reliability under elliptical contact elastohydrodynamic lubrication (EHL), an intelligent reliability assessment method is proposed. Contact stress under EHL is obtained by the mapping of oil film pressure, gotten by finite difference method (FDM), in the Hertz contact zone of the finite element model of rolling bearing. Considering the randomness of the EHL, material and fatigue strength correction factors, the limit state function is established by using artificial neural network (ANN). For finding the optimal reliability index and the design point, genetic algorithm (GA) based on normalized real number encoding is employed and the two adjusting factors are introduced into fitness function to resolve convergence and stability problem. Reliability sensitivity analysis is achieved by the advanced first-order secondmoment (AFOSM) method. Compared with the traditional Monte Carlo method (MCM), the proposed intelligent assessment method could embody the influence of EHL on contact fatigue reliability and has higher calculation efficiency and a wonderful global search capability in the whole optimization room.