This paper is concerned with investigating the application of Bayesian inference to monitoring the condition of turbojet engines. In order to overcome the typical necessity of a large number. of measurements for a deterministic solution of the problem, a statistic approach to the identification of maifunctions or degradation of engine components has been developed by combining a Gas Path Analysis model of the engine with statistical inference. In this way it becomes possible to infer the engine condition with a high level of confidence from the limited data provided by the sensors usuaIly available in aeronautical applications. The database for testing the method is generated by the same engine mode1 used in the identification, with the addition of random noise and offset errors to simulate measurement uncertainties and sensor malfunctions. The performance of the identification procedure is analyzed for several choices of the available me-as urements and sensors accuracy. The results illustrate the influence of these factors on the capability of identifying degradations or malfunctions of engine components and sensors.