If you are considering the net benefits of reducing the prospects of death using a costly technology then it is necessary to work out the value of a statistical life – a life that is anonymous but real. Non-economists find this macabre but the question makes perfect sense if budgets are limited as they always are. One difficulty is that what people reveal they do pay to reduce risks of death – revealled preference valuations – imply much lower values for life that valuations based on what people state their values to be – stated preference techniques. What values for life should therfore be assigned? Kip Viscusi sorts out these issues in a valuable survey.