[en] There is a continuing commitment in Canada to perform research to improve our knowledge of fission-product behaviour, and to understand the consequences of possible nuclear reactor accidents. This work was given additional justification by the accident at the Three Mile Island II nuclear generating station. During this accident, the release of radioactive iodine to the environment was low compared to the predictions of the Reactor Safety Study (WASH-1400), a study known to be overly conservative in some areas. There has been considerable effort, through both experiments and computer modelling, to re-evaluate the technical bases for predicting fission-product releases during nuclear reactor accidents. The research work is conventionally referred to under the general title of 'source-term' research. The source term is the amount and type of radioactive material that escapes the boundary of a nuclear reactor containment building, and thus could result in possible exposure of the general public. The source term depends on the type of reactor and the accident sequence. In general, fission-product release to the environment depends on the the airborne fission product concentrations within the containment building. This, in turn, depends on all the physical and chemical factors controlling the release from fuel, migration and transport, chemical speciation, surface interactions, and aerosol behaviour. The scope of the factors influencing the behaviour of fission products within the containment is sufficiently broad to include virtually all the research relevant to reactor safety. It includes studies of thermalhydraulics, fission-product releases from fuel under transient conditions, aerosol behaviour, fission-product chemistry, gas/solution partitoning and fission-product/surface interactions. This report will summarize four areas: (1) the use of the source term in the licensing of nuclear power reactors; (2) the methods used to determine the source term; (3) problems in source-term determination that require further research; (4) current research programs aimed at improving estimates of the source term