Condition based maintenance is a widely used maintenance strategy in modern civil aero engines. Engine health assessment and life prediction provide the essential information for making maintenance decisions. Firstly, principal component analysis method (PCA) was used to fuse multiple sensors data into a composite health index (CHI), and the reference models were trained by particle swarm optimizing support vector regression (PSO-SVR). Then the in-service sample CHI data was substituted into the reference models to do prediction. Finally, remaining useful life (RUL) of the in-service engine was calculated based on the similarity principle. The results show that the root mean square error (RMSE) of the in-service engine RUL prediction is 9.103, which proves the effectiveness and accuracy of the method.