Antineutrino emission in nuclear reactors arises from the beta decay of the neutron-rich daughters produced in heavy element fissions. The average fission is followed by the production of about six antineutrinos that emerge from the core isotropically and without attenuation. Thus the high fission rate - 10**21/second in a typical power reactor - implies a similarly high antineutrino rate. The high antineutrino flux near reactors compensates to a degree for the low interaction probablity of the antineutrino, making possible high statistics measurements in detectors of tractable size. This fact is a necessary condition for use of antineutrino detectors as safeguards and monitoring tools.

Aside from reasonable event rates (a few thousand per day in a standard configuration) reactor antineutrinos have a second feature that is particularly interesting from the standpoint of safeguards. The antineutrino rate, when normalized to the reactor power, is correlated with the plutonium and uranium content of the core. This correlation, known as the "burnup effect", may be used to track changes in the reactor's fissile inventory while the reactor is operating. As a reactor core proceeds through its irradiation cycle, the mass of each isotope varies in time. Initially, only uranium is consumed by fission. As the cycle proceeds, plutonium gradually builds up and begins to fission. Because the number of antineutrinos per fission differs between isotopes, the variation in the fissile content of the core is reflected in the antineutrino rate measured in a nearby detector. This correlation is such that at constant power, the antineutrino rate will vary by some 5 to 10 percent of its initial rate as the reactor proceeds through its normal burnup cycle. The change in rate is directly correlated with the ingrowth of a few hundred kilograms of plutonium. As demonstrated by the Chooz experiment in France, the absolute antineutrino rate can be measured to 3% precision including all systematic errors. Measurements made relative to reactor startup can be made even more precise, since many systematic errors are absent from such measurements.