Abstract: A method and apparatus for obtaining an accurate measure of a transfer function of either a linear or the linear part of a nonlinear system. This is accomplished by utilizing an estimation technique using a "periodic-random" stimulus consisting of a random stimulus of a finite record length that is repeatedly applied to the system under test until the stimulus and the system response become essentially periodic. Once periodicity is achieved within the desired accuracy, a first estimate of the transfer function is derived from the measured spectrums of the stimulus and the system response. Another "periodic-random" stimulus, uncorrelated with prior records, is then generated and the above procedure is repeated to achieve an average value of the transfer function within the desired accuracy.