This paper describes the development of a stochastic rail wear model. Once validated, the model can be used to assist in the strategic assessment of railroad track funding needs. The algorithms used for simulating rail wear use Markov processes, and the resulting transition probability matrix defines rail wear progression, as opposed to the more familiar regression-type model popular with engineers. The New Zealand railroad track database contained 10 years of rail wear data from which to develop and validate the model. The transition probability matrices for use in the model were developed with the first 5 years of the historical rail wear data, with the remaining 5 years set aside to validate the model. The development of the transition probability matrices is reported in the paper, together with the development of the initial condition distributions ready for use in validation of the model.