Tower cranes productivity are leading the construction jobsite duration. As a consequence, cranes are used by the owners very intensively. Tower crane structures are essentially made of steel plates or beams connected by welding and due to this intensive workload, fatigue design lifetime is therefore an important aspect to consider. Usual standards for fatigue design of tower cranes consider deterministic safety factors. This classical method leads most of the time to non-optimized structures with unknown and non-uniform safety margins. The proposed work takes place further to two projects called DEFFI (Reliability Approach in Fatigue Design for the Industry) and APPRoFi (Probabilistic Approach for Robust Fatigue Design). The goals of these projects were to experiment and to develop reliability approaches for the mechanical design against fatigue of industrial applications. Based on these previous works, the aim of this work is to quantify and optimize the safety margins with respect to fatigue of steel structures, by means of probabilistic approaches. To assess the reliability of crane structures, a probabilistic approach is used in order to estimate the probability of failure of each part of the crane. For tower cranes, the fatigue damage comes from the variations of stresses due to the combination of four variables: the load lifted by the crane, the radius of the load along the jib, the angle of orientation of the jib and, the slewing moment. The question herein is to determine the real impact of each parameter on the reliability of the structure related to fatigue lifetime. In this study, it is proposed to apply a probabilistic approach in order to evaluate the reliability of structural crane members and to perform a sensitivity analysis to understand the effect of each random variable on the reliability of structures.