A new algorithm for estimating power frequency is presented. Unlike conventional algorithms, the proposed algorithm is based on the fact that the magnitude gains of cosine and sine filters become different when the power frequency is deviated from the nominal value. This makes the algorithm capable of providing an accurate and fast estimate of the power frequency. To demonstrate the performance of the developed algorithm, various computer simulated data records are processed. The algorithm showed a high level of robustness as well as high measurement accuracy over a wide range of frequency changes. Moreover, the algorithm was highly immune to harmonics and noise.