In policy applications, industrial economists are wont to invoke the Kaldor–Hicks potential compensation principle to justify the use of deadweight loss as a measure of the welfare cost of market power. This usage rests on two assumptions. One of these assumptions, that changes in consumer and producer surplus are weighted equally, is well understood. The other assumption, that the marginal utility of income is constant, receives less attention. In a simple model, I show that if there is decreasing marginal utility of income, the use of deadweight loss as an index of market performance rests on shaky ground.