The radio altimeter system is an important part of the aircraft navigation system, the system failure will directly affect flight safety. Therefore, to ensure the safe flight of the aircraft, a fault diagnosis method based on fault trees and Bayesian networks was proposed in this paper. Firstly, starting from the system composition and working principle, taking the typical failure of the system ‘loss of radio altitude displays’ as an example, the fault trees were set up to comprehensively analyze the cause of the fault. On this basis, the built-in fault tree models were used to build the Bayesian network model, and to determine the maximum possible paths and key nodes of the fault through reverse reasoning to achieve the fault positioning. The outcomes demonstrate the excellent accuracy and dependability of the proposed method.