Despite the wealth of studies on the dynamic characteristics of peripheral auditory neurons, very little has been reported on the higher statistical moments of the neural spike train. The notable exception is the study by Teich and Khanna (1985) where both the mean and the variance of the neural count are reported. The simplest model one can ascribe to a neural spike train is a homogeneous Poisson process. However, experimental data do not bear out such predictions. Other models have been proposed but the general consensus is that the underlying process is far from simple. We offer an alternative account of the fluctuations that occur at the peripheral level. Our explanation does not rely on assumptions regarding the process underlying individual spikes. Instead, we make use of the information-theoretical model of the neuron that we have been developing over the past 10 years (Norwich and Wong, 1995; Wong, 1997). The two key results predicted by the model are that (a) the mean-variance ratio has an approximate value of 2 and (b) the distribution governing the neural count is Gaussian to a good approximation.