This paper proposes a statistical model for the characterisation of longitudinal road topography. The model is based on the Gaussian and offset Rayleigh distributions. The model originates from a new method to analyse and interpret the random nature of road profile data. The method was initially developed as a tool to assist in the simulation of shock and vibrations related to road transportation. The approach uses the double derivative of road profiles (spatial acceleration) as the main descriptor for road surface irregularities and applies a separate treatment to the transient and steady-state components of the signal to reveal their statistical nature. The statistical distributions of both components were found to correspond reasonably well with the offset Rayleigh distribution, thus enabling the characterisation of entire road sections or networks with three or five parameters.