This paper proposes a new theory of rational choice. Contra standard decision theory, this paper roughly argues that an agent should rank her choice options (in terms of how choiceworthy they are) according to their expected comparative utility. For any number of alternative options, a, b, c, d and e, the expected comparative utility of a is the probability-weighted average of the differences in utility between a and b, c, d or e across the various states – that is, for each state, the difference in utility between a and whichever of b, c, d and e carries the greatest utility in that state. In a number of ordinary decision cases, ECU Theory gives different verdicts from those of EU Theory.