The physics of nuclear reactions embodies essential information for the design, operation and decommissioning of nuclear systems, with applications spanning energy, safety, medicine, science, security and a great many other industrial processes. The types of knowledge required are as diverse as the applications, including on reaction probabilities (cross-sections) for many different types of reactions, emitted particle probabilities/energies/angles, probabilities for different fission fragment formation, and decay processes and their emitted particle data. Many of these data vary with the incident particle energy, potentially by factors of one million or more. In addition to the above requirements, it is important to have correlated uncertainties for all data in order to quantify and propagate uncertainties in simulations.

A general-purpose nuclear data library must therefore contain all of this information. It is the role of the nuclear data evaluators to craft databases of information for all of the elements and isotopes that may be required. Through the Nuclear Energy Agency (NEA) international EXFOR database, evaluators have access to a wealth of experimental data to guide this process, but only a small fraction of a library can be directly compared with measurements as a result of the extensive requirements for a general purpose library. Models and computer codes are developed not only to fill these gaps, but to ensure consistency of the data. Because of the fundamental challenges presented by the different aspects of relevant nuclear physics, there is no single model or code that can calculate more than a fraction of the required data. As a result, many models are in use within every major nuclear data programme, of which there are several that currently exist around the world.