Differences between estimated willingness to accept compensation (WTA) and willingness to pay (WTP) that are larger than can be explained by standard economic theory raise questions about which measures should be used for benefit–cost analysis (BCA). These differences do not create a new problem but accentuate an existing one: the fact that the Kaldor–Hicks compensation test is ambiguous when its two components conflict. This conflict is more likely when the difference between WTA and WTP measures of a change is large. In many cases, the same individuals receive benefits and incur costs from a policy change and their preferences for the policy cannot depend on whether they ask whether their WTP for the benefit exceeds the cost they will incur or their WTA to forgo the benefit exceeds the cost they will save. In cases where benefits and costs are incurred by different people, it seems more useful to evaluate the fundamental question – whether the benefits to some justify the harms to others – than to obscure this question through a technical debate about valuation measures.