Utilitarianism plays a central role in economics, but there is a gap between theory, where utilitarianism is dominant, and applications, where monetary criteria are often used. For applications, a key difficulty is to define how utilities should be measured and compared. Drawing on Harsanyi’s (1955) approach, we introduce a new normalization of utilities ensuring that: (i) a transfer from a rich population to a poor population is welfare enhancing, and (ii) populations with more risk-averse people have lower welfare. We study some implications of this “fair utilitarianism” for risk sharing, collective risk aversion, and the design of health policy.