The higher our aspirations, the higher the probability that we have to adjust them downwards when forming more realistic expectations later on. This paper shows that the costs induced by high aspirations are not trivial. We first develop a theoretical framework to identify the factors that determine the effect of aspirations on expected utility. Then we present evidence from a lab experiment on the factor found to be crucial: the adjustment of reference states to changes in expectations. The results suggest that the costs of high aspirations can be significant, since reference states do not adjust quickly. We use a novel, indirect approach that allows us to infer the determinants of the reference state from observed behavior, rather than to rely on cheap talk.