This paper introduces a novel modelling and identifying technique for the transformer internal faults. Transmission line method (TLM) is used for the modelling transformer internal faults with a linear and non-linear inductor model. Discrete wavelet transform (DWT) is used for identifying fault currents. The proposed techniques are simple to implement and provides a reliable and computationally efficient tool. Results for a range of simulated faults are used to compare the efficiency of DWT with conventional Fouier Transforms. It is shown that it provides a more reliable fault discrimination algorithm than conventional methods.