[en] By the end of 1998, 992 assemblies had been irradiated since the first loading of MOX fuel assemblies in French PWRs in 1987. 20 reactors are now authorized to use MOX fuel. Operating MOX fuel in PWRs up to an average discharge burnup of 37 GWd/t did not cause any particular problem and MOX fuel reliability, based on increasing experience feedback, is as good as the UO₂ one. In parallel, a considerable experience on MOX fuel performance has been accumulated by post-irradiation examination of rods irradiated up to more than 50 GWd/t (4 cycles) in PWRs (surveillance programme) and in experimental irradiation programmes. Ongoing R and D at the CEA, in collaboration with the EDF and FRAMATOME, is directed towards gaining a better understanding of MOX fuel behaviour in normal and off-normal conditions in order to improve and validate the models used to justify fuel behaviour prediction calculations and to support the licensing of extended burnup. Recent results in PWR experience feedback as well as R and D programmes will be discussed. As MOX fuel has now reached a sufficient maturity, the next step is to achieve parity between UO₂ and MOX fuel. This involves a single management strategy for UO₂ and MOX assemblies with an annual quarter core reload type. The goal is to reach a maximum assembly discharge burnup of 50 GWd/t for MOX and UO₂ fuel assemblies. The new FRAGEMA AFA 3G structure will increase margins in fuel rod design with regard to the end-of-life internal pressure criteria. The objective of obtaining a parity between UO₂ and MOX must be achieved even beyond the initial goal of 50 GWd/t for maximum discharge burnup. This is why another programme was launched with the purpose of substantially improving MOX fuel assembly discharge burnup, i. e. up to 70 GWd/t, as has been the case for advanced UO₂ fuel previously. The R and D programme mainly focuses on increasing fission gas retention, which is the main obstacle to reaching very high burnup. The improvement of fuel assembly design will also be integrated to increase margins. The main steps of the programme are the following: optimization of pellet microstructures and validation in experimental reactors, build-up of experience feedback at elevated burnup in commercial reactors, both for current and experimental products, adaptation and qualification of the design models and tools. The product resulting from this new development could be qualified around 2010. (author)