Geological disposal of high-level nuclear wastes (HLNW) is planned over hundreds of thousands of years, or even up to several million years. How can we predict the behavior of materials over such durations and, in particular, their corrosion?

We propose a global approach which can be implemented for the geological disposal of high-level nuclear wastes (HLNWs), and present this alongside the main corrosion-related issues. This approach was formalized by Digby's participation at the beginning of the 2000s. Some specificities of the French concept of a geological repository will be underlined: clay underground repository, reversibility concept, HLNWs mainly composed of a glass matrix, a carbon steel overpack, etc.

The main approach is based on “four pillars” which we have developed and which include:

• The use of experimental data obtained in laboratories for initial estimation of service lifetimes and corrosion mechanism investigations.
• Development of modelling (mechanistically based and also stochastic modelling) based on data and mechanisms found in laboratory experiments for robust and reliable prediction.
• The use of archaeological artefacts to provide data for long-term corrosion rates for the investigation of corrosion mechanisms and for validation of corrosion modelling.
• The importance of integrated experiments in underground laboratories is underlined to be sure that all the interactions are integrated.

The whole approach is iterative and includes the integration of the evolution of knowledge.