How Should We Revise Our Beliefs about Nuclear Power Safety after Three Mile Island?

In light of the accident at Three Mile Island, the paper presents a preliminary analysis of the compatibility between the analytical work of the Reactor Safety Study (Rasmussen Report) and actuarial experience. The technique is a “macroanalytic” approach rather than the “microanalytic” method of the Reactor Safety Study.

The first question asked is how likely it is that an accident as severe as that at Three Mile Island would occur if the Reactor Safety Study is correct. Using the most likely estimates, it is concluded that the chances are 1 in 80 that such an accident would occur this soon, but given uncertainties about parameters it might range from 1 in 17 to 1 in 625.

The second question addressed is how we should revise our estimates of the safety of nuclear power given the Three Mile Island experience. Using the technique of maximum likelihood, our best guess estimate of the risk of accidents causing at least one fatality rises from the Reactor Safety Study’s 32 per million reactor years to about 2000 per million reactor years.