Although the lifetable methodology is a standard tool in epidemiology and risk assessment, there are a number of differences in the way it has been applied by various advisory committees that have attempted to estimate radiation risks. The most fundamental of these differences concerns the choice of parameter to be estimated: the "excess lifetime risk" is the difference in lifetime risks between exposed and unexposed populations; the "risk of exposure-induced death" is the lifetime risk of dying of a disease attributable to exposure. These two quantities are not the same, even at low doses. Although both quantities have some utility in risk assessment, the "risk of exposure-induced death" comes closer to capturing the total impact of exposure. Other differences between reported risk estimates include details of the calculations, the baseline rates and age distributions of the exposed population, the forms of the models for excess rates, handling of organ-specific doses, and the groupings of cancer sites. These issues are discussed theoretically and illustrated with comparisons of the BEIR V and UNSCEAR reports. Although the risk estimates from these two reports are similar for most cancer sites, it is shown that this happens to be the result of an approximate cancellation of a number of differences that could be quite large.