Up to now， the problem of single-phase-to-ground fault location has still not been satisfactorily solved. In analyzing the laws governing the zero mode and aerial mode traveling wave transmission， the authors have found the relationship between the transmission time gap of the zero-aerial mode traveling wave and fault distance， and the law of variation of Lipschitz index of the zero mode wave head with fault distance. Based on the aforementioned analysis， a wavelet transform and BP neutral network based fault location algorithm is presented， which takes the advantage of the time-frequency analyzing capacity of wavelet and the robust nonlinear fitting capacity of neural network. The effectiveness of the criterion has been verified by the ATP simulation results， indicating that the algorithm proposed is immune to such factors as branch lines， primary phase angle of fault and transitional resistance.