Internal combustion engine (ICE) is a complex mechanical system. It is difficult to identify ICE health status for lack of fault data and its complex vibration characteristics. In order to effectively evaluate the health status of ICE, a fault diagnosis method based on principal component analysis (PCA) and support vector machine (SVM) is investigated. Firstly, principal component features of ICE are extracted through eliminating redundancy and reducing the dimension of original vibration signal feature parameters by PCA. Then, these features are taken as training samples and one-against-all SVM classifier is designed to identify health status of ICE by using radial basis kernel function. Through analyzing the vibration features of ICE under different conditions, experimental results indicate that the fault diagnosis method can effectively recognize different status of ICE.