Accurate predicting of failure rate of power transmission equipments can enhance credibility of reliability assessment for transmission systems and its prediction results, so as to improve the reliability management level of power transmission system. However, due to the complexity of power transmission equipments distributed geographically, and randomness and diversity of operation states, the failure rate of power transmission equipments have characteristics of time-varying and stochastic. As it is difficult to obtain the information reflecting equipment faults, conventional methods of regression analysis and decomposition analysis and time series analysis could not be used to establish accurate prediction model of failure rate. This paper presents a predicting method of transmission failure rate based on autoregressive-moving average(ARMA) model in order to adapt to time variation and randomness of equipment failure rate of power transmission system. The method is focused on equipment failure rate of whole power transmission system without the difficulty of obtaining information for reliability evaluation in conventional analytic methods, and less information is needed. So the validity and accuracy of reliability evaluation for equipments in whole transmission system are improved effectively. Finally, the method is used to predict the failure rate of a provincial 220kV power transmission system. The results have shown that the method has high accuracy, and can be applied to engineering practice to improve stability of power systems .