A new numerical technique for evaluating power system frequency from either a voltage or current signal is presented. The technique uses discrete time values of the input signal, taken at a fixed sampling rate, to provide an estimate of power system frequency accurate to within typically 0.001 Hz. The technique is shown to be capable of tracking frequency under dynamic power systems conditions and is immune to the effect of harmonics. It is further shown that the algorithm can be easily adapted to process three-phase signals such that the positive phase sequence component can be utilised thus increasing the reliability of the measurement under fault conditions.