Predicting crack initiation life (CIL) of a mechanical component or a structure in service remains difficult since the crack formation process is of stochastic nature. To ensure a high level of safety and reliability, it is essential to have an appropriate probability distribution law of the CIL to ensure that cracks can be detected before reaching a critical length. In the present study, a stochastic model is used to predict the number of cycles corresponding to the formation of a crack 500 μm long resulted from the nucleation, growth, and coalescence of multiple microcracks. The model is applied in the case of a 316L austenitic stainless steel for different plastic strain ranges.