This paper describes a probabilistic damage tolerance concept for the fatigue damage in gas turbine engine components. The statistical scatter in time to crack initiation is considered using a Weibull distribution, while the stochastic crack growth is treated by a Markov chain approximation. The approximation is a unit-jump discrete state/discrete time Markov chain where damage states are related to the fatigue crack depths. The model parameters are estimated from experimental statistics and a linear fracture mechanics simulation of the fatigue propagation. The effect of in-service inspection is considered by the use of Bayesian updating techniques. The analysis verifies that it is possible to extend the service life and maintain the fatigue reliability of critical components by an appropriate periodic inspection program.