Frequency is an important parameter for the power quality analysis. When two or more adjacent spectral lines are too close, many spectrum estimation algorithms fail to distinguish these frequency components. A modified high-resolution Singular Value Decomposition (SVD) method for power quality signal analysis by using down-sampling technique is proposed in this paper. With adopting down-sampling technique, a scaling factor is introduced to separate the spectral lines from each other, and then the correct spectra can be estimated. The performance of the proposed method is validated by testing the actual measured signal. Results are compared with those obtained from several FFT-based and SVD methods, and the commercialized power quality meter. It shows that the proposed method can precisely detect the frequency components of the measured power signal with a high resolution.