The INES scale

The scale

The right-hand column counts only the 19 events documented on this site, which are the significant ones. It is not a count of all events ever rated: national regulators rate events at levels 0 to 2 routinely, and those run to hundreds a year worldwide.

Three things the scale is measuring at once

A rating is the highest of three separate assessments, which is why events with very different characters land on the same number.

• People and the environment. Radiation doses to workers and the public, and release of material off site. This is the criterion that produces the levels above 4.
• Radiological barriers and control. Damage to the fuel, contamination inside the facility, loss of control over a source. An event can be rated here with nothing at all released to the outside world.
• Defence in depth. How much of the safety margin was consumed. Nothing has to go wrong for this criterion to apply. If every layer of protection but one had failed, that is rated even though the plant behaved correctly in the end.

Davis-Besse is the clearest illustration. Boric acid ate through almost the entire thickness of the reactor vessel head, leaving a stainless steel liner a few millimetres thick holding back the primary coolant. Nothing was released, nobody was hurt, and the unit was rated 3 entirely on the third criterion.

What the scale does not do

• It is not a measure of deaths. Chernobyl and Fukushima Daiichi are both rated 7 and their death tolls are not comparable. The accident pages here state prompt deaths and long-term attribution separately, because collapsing them into one number is how the argument about nuclear safety usually goes wrong.
• It is not linear. Each step is roughly an order of magnitude, so the distance from 5 to 7 is not "two worse", it is about a hundredfold.
• It is not applied consistently across countries. Ratings are assigned by national authorities, sometimes provisionally and sometimes revised months later. Fukushima Daiichi was initially rated 5, then raised to 7.
• It did not exist for most of the history it is used to describe. INES was introduced in 1990. 13 of the 19 events on this site happened before that, so their ratings are attributions made afterwards by others rather than assessments any regulator issued at the time. Chernobyl, Kyshtym, Windscale, Three Mile Island and the rest of the pre-1990 record are all in that category, and the tables here mark them.
• It says nothing about cost or about how long a site stays closed. The economic consequence of an event and its INES level are not related.

The rated events on this site

Each has its own page with what happened, what it changed, and where the figures come from. The mechanism behind several of them is covered under decay heat, and the reactor families involved under moderators and coolants.