A stochastic model for the accumulation of fatigue damage in composite laminates is proposed. The model considers the accumulation of fatigue damage as an evolutionary stochastic process and attempts to predict the distribution of damage in terms of the stiffness loss in the laminate as a function of the number of load cycles. The cumulative damage process is modeled as an embedded discrete time, finite state, non-stationary Markov process utilizing a stationary discrete time, finite state, stationary model in conjunction with a time transformation-condensation method. Furthermore, the mechanisms of the complex damage development process in composite laminates are incorporated into the model. This feature affords a more complete and accurate representation of the fatigue damage accumulation process in which the states of the model can be directly related to the stiffness loss in the laminate.