A method is described for studying the dynamic properties of unit responses using sounds which are amplitude modulated with pseudorandom noise. It is based on computation of cross-correlation functions between the signal modulating the stimulus and the neural discharge frequency. The discharge frequency is represented by a cycle histogram of the neural discharges locked to the repetition of the noise. Transfer functions of the system under test are obtained on the basis of the spectrum of the computed cross-correlation functions. The results obtained using this method in study of the dynamic properties of the responses of single units in the cochlear nucleus of the rat were compared with results obtained using sounds modulated with sinusoids. The dynamic properties were described by gain functions showing the ratio between the degree of modulation of the neural discharge frequency and the degree of amplitude modulation of the tone. The results using these two methods were similar. It was furthermore found that the reproducibility of the results over several hours was good. It is concluded that the responses of these units can be described using linear system theory when the modulation depth is kept low. In many units the dynamic properties become a function of the sound intensity. A transfer function may thus only be a valid description of the dynamic properties of a unit at a certain stimulus level. The advantages of using random noise signals in studies of the dynamic properties compared with sinusoidal modulation signals are discussed.