Winding insulation is an important constituent of a transformer. Any weakness of insulation may result in winding deformation and hence, in the failure of the transformer. Insulation failure arises due to over voltages, notches, glitches, etc. Comparison of recorded neutral currents is a standard technique used for the detection of fault in the power transformer winding insulation during impulse test. Any shift in the recorded waveforms between reduced and full voltage confirms the existence of fault. In the present work, the recorded neutral current waveforms obtained from a 61MVA, 11.5/230 kV generator transformer have been considered for analysis. Faults have been created in the discs of the High Voltage windings at specific locations. An impulse voltage of 100V is applied to the high voltage winding and the neutral currents for the healthy case and the faulted discs windings have been recorded using digital oscilloscope. These recorded neutral current waveforms are analyzed using wavelets for fault identification. The component of the noise which is inherent in the signal during recording was isolated using Daubechies7 wavelet and these denoised signals were analyzed further using Daubechies7 wavelet to identify the fault in the winding.