The safety advantage of next-step fusion machines, such as ITER, and o
f future fusion power reactors, comes not so much from the engineered
safety features incorporated into the design, as it does from the inhe
rent, or intrinsic, features of the machine/reactor. The key intrinsic
feature is the small quantity of radioactive fuel (tritium) and its l
ow health hazard. The radioactive health hazard is small enough that e
ven the largest unmitigated release would produce offsite consequences
that are small. Nevertheless, engineered safety features are necessar
y to reduce the probability of an unmitigated release to a small enoug
h value, such that the resulting risk is within risk targets. The risk
targets are established such that the total accident risk is not grea
ter than the risk from normal operation. Risk targets are also useful
in determining the performance requirements for the engineered safety
features. Such a risk-based approach should be considered when develop
ing fusion safety requirements, taking account of the low radiation ha
zards involved.