Abstract. Achieving the economically viable release of energy by the fusion process has proved a major scientific and technological challenge. The nuclear reactions involved in the fusion process are well understood, but the interaction of the neutrons with the materials surrounding the plasma is an area of current research. The neutrons have energies larger than those experienced in fission and this has been a driver to extend the types and amounts of nuclear data. Designs for fusion devices require neutron and photon transport calculations to generate the spectra required to provide the nuclear responses. Such calculations require cross section data for all elements present in the device. ITER is the next step on the road to demonstrating the commercial generation of electricity. The importance of nuclear data for ITER was realized many years ago, which led to the release of FENDL-1 in 1995. This contained both general and special purpose files; an example of the latter is activation. Development of these files focused on completeness -including γ production and covariances; the files generally had an upper energy limit of 20 MeV. The types of data in the fusion relevant libraries are sufficient for all foreseeable fusion devices. However, the transition in fusion research from plasma physics towards technology requires an understanding of the materials damage caused by the intense fluxes of neutrons that will be found in power plants. To gain this knowledge a materials test facility, IFMIF, is required. This is planned to operate in parallel with ITER and will use accelerated beams of deuterons striking a flowing lithium target to generating an intense neutron field. The neutrons will mostly be fusion relevant, but there will be a high-energy tail extending up to ∼55 MeV. These high energy neutrons further extend the amount of nuclear data required and have stimulated much theoretical and experimental work covering the energy range >20 MeV. Examples of recent general purpose evaluations, new activation libraries and benchmarking are described and the challenges for nuclear data for fusion discussed.