AbstractNext-generation fusion devices may consume several kilogrammes of tritium per year [1]. The supply of tritium for fusion depends upon the availability of man-made sources, and the size of the fusion and commercial demands.The most significant uncommitted tritium resource is the 2.2–2.3 kg/y from the Ontario Hydro Darlington facility. Military tritium production is a large source, but is almost entirely committed to military programs, except for about 0.1–0.5 kg/y made available for commercial uses. Side production of tritium in power reactors is possible, but limited in practice. It is estimated that an additional 0.2–0.5 kg/y by the year 2000 is a reasonable assumption from this and other sources, such as fuel reprocessing facilities and the Chalk River tritium removal facility. Also over the next decade, fusion demand will average about 0.01 kg/y, while commercial demand will be roughly 0.2 kg/y and possibly grow to 1 kg/y.The effect on a fusion engineering test reactor starting up around the year 2000 is that the machine operating lifetime may be limited, or some tritium breeding required. Some scenarios are quantified here. Assuming for example, about 2 kg T/y is available for fusion use starting 1988, then a 700 MWf machine starting in 2000 would need a 70% tritium breeding rate to support a ten-year testing phase at 30% availability (i.e. 3 MW-y/m2 fluence at 1 MW/m2 wall load).