We introduce the Cautious Utility model and show that it provides a new approach to three phenomena at the core of behavioral economics: the endowment effect, loss aversion, and violations of Expected Utility due to the certainty effect. In our model, all three phenomena stem from uncertainty about which utility to use and caution. We show how this model can help organize empirical evidence, some of which is incompatible with leading alternatives, and is both conceptually and behaviorally distinct from other popular approaches.