Abstract: The equipment reliability data processing is the basis of reliability centered maintenance (RCM) and life cycle management (LCM) in nuclear power plant. However, in actual failure data processing, the problems such as small-sample and non-independent data caused by maintenance are put forward. To resolve the problems, a processing method combined double-parameter Weibull distribution as the life model and Bayesian method for small samples was proposed, and was validated using actual nuclear power plant operating data. The results show that the processing method has better applicability and accuracy to deal with the situation of small samples and the problems of repairing and aging in nuclear power plant.