Improvements in nuclear safety are often achieved through introducing a new safety measure that reduces the frequency of a hazardous accident rather than its consequences. To carry out a J-value analysis, it is necessary to calculate how a reduction in accident frequency extends the life expectancy of the potentially exposed group of people. The paper presents two methods for calculating the loss of life expectancy associated with accidents of a certain severity occurring with a defined frequency. The first begins by using an equivalent, prolonged radiation exposure to represent the effects of the accident occurring once per year over the given period of operation. The resultant loss of life expectancy is then scaled by multiplying by the frequency of occurrence. The second method calculates the loss of life expectancy brought about by a single accident occur-ring during the given period of operation and scales this by multiplying by both the length of the operational period and the frequency of occurrence. Results derived using the first method show I hat there is a relatively small effect on loss of life expectancy per accident if several accidents are assumed to occur during a typical period of operation. This conclusion permits a simple assessment of the effect of possible, multiple accidents. The accuracy of the second method is found not to be compromised materially by ignoring the possibility of multiple accidents. The second method is shown to be slightly more conservative than the first, and also somewhat more accurate. Calculations of the loss of life expectancy may be carried out before and after the new safety improvement has been implemented, and the difference between the two results will be the life extension brought about by the new safety measure. (C) 2008 The Institution of Chemical Engineers. Published by Elsevier B.V All rights reserved,