This paper develops state models for estimating the optimum performance of analog and computer impedance relays in the presence of noise due to non-fundamental frequency components. An Extended Kalman filter is used to determine the optimum estimates and error covariance of impedance magnitude and phase angle. It is shown that impedance relays have performance limits dependent upon the amount of noise present. The effects of sampling rate of input signals, data-window used for fault detection, and cut-off frequency of anti-aliasing filter on the performance of a computer relay are also investigated. Digital computer simulation results for both analog and computer impedance relays are presented. Copyright © 1984 by The Institute of Electrical and Electronics Engineers, Inc.