Abstract A good understanding of the mechanical behaviour of the reactor pressure vessel (RPV) lower head is necessary both for severe accident assessment and for the definition of appropriate accident mitigation strategies. Indeed, a well-characterized failure of the lower head leads to a better evaluation of the quantity and kinetics with which core material can escape into the containment. These are the initial conditions for several ex-vessel events such as direct heating of the containment or molten core–concrete interaction. In this context, the objectives of the joint on-going work of the WP10-2 group of SARNET are: (1) improvement of predictability of the time, mode and location of RPV failure; (2) development of adequate models with the ultimate aim of being included into integral codes; (3) interpretation/analysis of experiments with models/codes combined with sensitivity studies; and (4) better understanding of the breach opening process in order to better characterize the corium release into the containment. Different approaches are considered: a simplified but well predicting model recently implemented in the severe accident Astec and Icare-Cathare codes, and viscoplasticity models implemented in the Cast3m, Ansys and Code_Aster finite element codes. Several failure criteria are considered: stress criterion, strain criterion and damage evaluation (coupled way or post-evaluation). In this paper, the OLHF-1 experiment has been used to assess the models, to perform sensitivity studies and to evaluate failure criteria that could be applied in the case of reactors. All the partners performed 2D axisymmetric analyses, allowing the evaluation of time, mode and location of vessel failure. Nevertheless, CEA conducted further 3D calculations in order to study crack propagation and the corresponding results will be presented separately at the end of the paper. The numerical formulation of the different models used is given and a comparison of experimental and numerical results is presented. The paper also shows the progress made with the objective of defining failure criteria that can be used for reactor vessel applications.