Abstract: In order to overcome the flaws of traditional fault tree analysis which can only describe two states: normal or fault, a combined method based on fault tree (FT) and Bayesian Network (BN) for landing aircraft veering off runway analysis was put forward. The new method not only took the advantages of logical analysis of fault tree but also the function of the description of polymorphic events and probability calculation of Bayesian network. In addition, a transform algorithm from FT to BN was proposed to solve the model. Based on the data about aircraft veering off runway during landing phase of Civil Aviation Administration of China from 1996 to 2010, the key factors causing runway excursion landing accidents of aircrafts were determined and then ranked by their importance. The result indicates that standing water, dysfunction of thrust reverser or airbrake, complex weather conditions, deficiency of piloting technology, jammed nose wheel steering, night operation or adverse influence from lights, failure of crew resource management, accumulated ice or snow should be the key concerns whose posterior probability are greater than 0.3; and targeted preventive measures should be developed toward the major risky factors. The analysis result produced by the model is consistent with the fact.