Arrow et al. revisit the case for using benefit-cost analysis in a developed country, the USA, where markets work reasonably efficiently and where the capacity to implement such studies is undoubted. Their recommendations deserve wholehearted support in that context, particularly their recommendation 1 calling for a comparison of gains and losses from regulatory actions. Those who have not worked in government will recognise that most decisions are not in fact made with any form of calculus that we might describe as 'cost benefit thinking'. Indeed, the whole process of policy priority setting is all too often ad hoc, reactive, crisis-based and over-responsive to often ill-informed pressure groups (of all kinds).