The fault process of the injector has been discussed according to the variation laws of its start signal wave, the diagnostic parameters are put forward, such as waveform amplitude, the width of rising edge and breadth of waveform . The diesel engine injector fault diagnosis model, which containes fuzzy neural network (FNN) and the genetic arithmetic (GA), is established based on fuzzy logic reasoning and working mechanical of the injector. The fault type is also established according to characteristic parameters of the injector start signal, and the fuzzy logic reasoning of diesel engine injector fault diagnosis is put forward. Based on different fault injector start signals and FNN-GA fusion algorithm, the injector fault model is diagnosed and identified, the on-line diagnostic strategy of diesel engine injector fault is put forward and the fault tests on drive current of the injector are carried out. The results show that the diesel engine injector fault diagnosis model is reasonable; the RBF-GA diagnostic strategy has the good resolving power and can be fitted forthe on-line diagnosis of the injector fault.