The following explanation focusses on magnetic confinement of deuterium-tritium-fuelled plasmas, such as those in ITER, but similar or even stronger arguments apply also to other fuel combinations and to laser fusion.

The fusion process is inherently safe.

Leak-tight confinement barriers are essential to produce fusion reactions. Equipment failure quickly leads to plasma extinguishment.

No chain reaction is involved and the reaction is thermally self-limiting.

There is no danger of a large jump in plasma power output, since normal operation is close to pressure limits which already maximise the number of fusion reactions that will occur. In ITER, because of experimental uncertainty, it is possible for the plasma to operate at somewhat (<1.2) higher power levels than planned, but these can be easily brought under control in a matter of seconds.

The fusion process is limited to a few seconds burn, without continuous refuelling.

Achieving low loss burn conditions is a delicate matter and requires many conditions to be satisfied - the failure or change of a single one enhances plasma energy losses and terminates the burn. Halting the fuelling quickly extinguishes the plasma. In ITER about 0.5 g of fuel is in the machine at any time, and the fuelling/exhaust rate is also about 0.5 g/s. Even if the exhaust fails, the plasma is quickly poisoned by impurities, and extinguishes.

The power and energy densities in the reactor and plasma are low.

The main sources of energy which can damage ITER are pressurised coolant, chemical reactions (e.g. of leaking coolant and hot materials, or of hydrogen and air), heat from the fusion reaction in the plasma, and magnetic energy in the coils. There are no large stores of chemicals or other energy sources able to cause powerful explosions. ITER is designed such that its hardware avoids the unexpected release from energy sources or mitigates the consequences of any such release to acceptable levels not only for the general public, to ensure the ultimate safety of the plant, but also for plant operators, to protect their investment. To help in these respects, ITER has large heat transfer surfaces and heat sinks which transfer and absorb energy, maintaining low temperatures and avoiding melting of components. The same will be true in a power reactor, but the margins needed for ITER should be able to be reduced, and the overall power density should be able to be increased.

The reaction products are either absorbed in surrounding structural or tritium-breeding materials (neutrons), or are non-radioactive (helium).

In ITER nearly all materials around the plasma are to shield the surrounding equipment, whereas in a power reactor the bulk will breed tritium from lithium-containing materials, ready to burn it in the plasma.

Activated structural materials from neutron irradiation are not mobile except dust and corrosion products which form only a small fraction.

The neutrons produce activated waste materials. Dust is formed by sputtering from high energy particles in the plasma hitting the surrounding material surfaces. Although not necessarily a problem itself, this dust can become contaminated with tritium. Coolant channels can become corroded, especially in high nuclear radiation fields, and the corrosion can dislodge and be freed if a coolant pipe breaks. In ITER the coolant chemical control system is capable of maintaining coatings of activated corrosion products well below 10 kg per loop, with less than 60 g as loose material or ions in the coolant (these limits are used in accident analysis). In a power reactor this aspect will be further optimised.

Negligible operational environmental impact.

The potential risk to the local environment is limited and is reduced as low as judged reasonably achievable by the independent nuclear regulator in the country concerned.

Negligible long term environmental impact.

Neither the provision of fuel or plant hardware, nor its removal after use, places an intolerable and uncertain burden on current or future generations.