Within the context of Probabilistic Safety Assessment (PSA) studies Level 1 (which deals with risk of core damage) performed on nuclear power plants, crane bridges failure have been identified as a contributor to the probability of core meltdown. Depending on the reactor type and on the age of the reactor design, one of the significant failure mode can be related to anchorage failure during earthquake. Therefore, the issue of dynamic behavior of crane bridges needs to be considered within the global framework of the safety demonstration of a plant. In this context, it is necessary to improve knowledge about the dynamic behaviour of this equipment in order to fully understand how failure would occur in the case of far beyond design loadings induced by most severe earthquakes (that are considered in risk assessment). In this view, it appears to be important to estimate accurately the forces transmitted to the anchorages. In addition, the incorporation of diverse sources of uncertainties through a fragility curve still raises several questions such as: what are the main variables to be considered as random? What are the failure criteria to be used? Does the hypothesis of a lognormal distribution remain justified for seismic inputs for which the intensity is beyond the design range?In order to provide answers to these questions, a benchmark endorsed by the Working Group on Integrity and Ageing of Components and Structures (WIAGE) of OECD/NEA/CSNI will start in 2020. The main objectives of this action are (i) to identify best practices to model seismic behavior of crane bridges; (ii) to identify relevant failure criteria; and (iii) to establish international consensus on the definition of seismic fragility of cranes.An experimental campaign on a scaled model of an overhead crane bridge was carried out in 2015 on the AZALEE shaking table of CEA in France and the results have been gathered in a large database. On one hand, some of these data will be used by participants to characterize and calibrate their models and, on the other hand, some other data will be used to assess the predictive capacity of the models. The benchmark will be concluded by a restitution workshop in which the different participants will be gathered to exchange and discuss their models and results they have obtained. In this way, best practices for modelling overhead cranes under seismic load will be identified. The workshop will provide a synthesis of lessons learnt and recommendations based on findings of the benchmark analysis. A proceeding report including documents submitted by participants is envisaged and the findings of the benchmark analysis will be summarized in the restitution workshop that will be held in 2021 and a NEA report in 2022.In this paper, an overview of the benchmark will be presented, including main objectives, test specimen, and detailed benchmark process.