A model for probabilistic growth of complex fatigue crack shapes that will be used to perform sensitivity studies and to optimize inspection intervals for railroad tank cars is presented. In this model, crack growth is based on results from a deterministic crack growth analysis with additional terms to introduce stochastic behavior. The family of complex shapes that cracks grow through is parameterized by a single degree of freedom, which simplifies the probabilistic model. Comparisons between experimental and numerical results suggest that the probabilistic model is capable of representing realistic crack growth using reasonable crack growth parameters, which will enable the sensitivity and optimization studies for which the model is intended.