Abstract: According to the traits of fault vibration signals from a diesel engine, a new method was presented for the fault diagnosis of the signals based on EMD(empirical mode decomposition) and FCM(fuzzy c-means clustering). The time series of vibration signals of three different air valve clearances (0.4, 0.6 and 0.75 mm) was decomposed via EMD, then the energy ratios in percentage of the first six intrinsic modal functions obtained by EMD were calculated separately, and the relevant energy ratios were taken as the characteristic parameters reflecting what state the fault is in. Those parameters were analyzed by FCM algorithm. Experiments indicated that the diagnosed results of all samples conform to actualities, i.e. the method can effectively diagnose the faults of air valve clearance of a diesel engine.