This paper presents a wavelet denoising method for partial discharge (PD) signal extraction from the electromagnetic interference (EMI) existing in electrical machines such as high frequency sinusoidal noise, pulse-shaped noise and stochastic noise. The proposed denoising method is performed through multiresolution signal decomposition, thresholding of wavelet coefficients and signal recovery. Two types of numerical simulating PD pulses immersed in severe noise have been investigated. The paper demonstrates that the proposed method has tremendous potential for extracting PD pulses from a noisy background, thus providing reliable signals, on the basis of which, assessments can be made of insulation integrity in high voltage plant