Recently, the mechanism and possible roles of high interspike interval variability in neurons has been the focus of much attention. This paper is concerned with a statistical method to analyse the response characteristics of a neuron stimulated by random interspike intervals. We present a new method of calculating the natural stationary distribution of the internal states, and use this directly to calculate the Lyapunov exponent of the neuron model. Our method completely avoids having to produce long random orbits for the calculations, with the Lyapunov exponents calculated directly using an integral (a space average). Further, we argue that random iteration is an inefficient and inaccurate way of performing statistical analyses.