Judgment of the performance of automatic control systems often involves a consideration of the behavior of such systems in the presence of random noise. The noise may be of an undesirable nature, such as receiver noise, or it may represent an ensemble of statistical inputs which the system is to follow. Analytical techniques exist for determining the response of linear systems to noise inputs. However, no general methods are available for analyzing nonlinear systems subject to noise inputs. An electric analog computer, together with a suitable source of noise, may be used to study nonlinear systems. This paper describes an electronic noise generator designed for analog computer or simulator use. Methods are discussed for measuring the important characteristics of low frequency noise, such as the mean value, spectral density, amplitude distribution, and autocorrelation. Particular attention is devoted to the length of time necessary to establish satisfactory estimates of the properties of low frequency noise.