Abstract: The blind source separation algorithm based on DWT-EEMD used in noise elimination of power line interference for magnetotelluric is put forward in this paper. This method takes advantage of the excellent characteristics of DWT, EEMD and blind source separation, and adopts blind source separation to eliminate the noise after DWT and EEMD processing. The main advantage of this method lies in the DWT-EEMD model adopted and the introduction of the adaptive weighting factor, in reducing uncertainty of restored signal amplitude by independent component analysis algorithm at the same time, it has made the original signals still be well separated in condition of power frequency interference noise amplitude much higher than original signal amplitude. By the real magnetotelluric signal processing, it is showed that this method makes apparent resistivity curve and phase curve both become smooth and steady, and it better eliminates the power line interference noise in magnetotelluric signal.