A general class of stochastic models for choice behaviour is described, which can be regarded as finite Markov chains with absorbing states and which subsumes some particular models discussed by Audley and Pike (1965). The equations for response probabilities and moments of response latency distributions are obtained from the properties of such chains and a very general approach is given in terms of ‘transition‐similar’ systems, which includes discrete time and continuous time processes. Generating functions for the systems are obtained and variations in parameters controlling the rate of occurrence of events in time are also considered. A brief discussion centres around some of the problems which arise in such a general treatment and the reasons for its usefulness.