The correct prediction of how alternative states of the world affect our lives is a cornerstone of economics. We study how accurate people are in predicting their future well-being when facing major life events. Based on individual panel data, we compare people's forecast of their life satisfaction in five years' time to their actual realisations later on. This is done after the individuals experience widowhood, marriage, unemployment or disability. We find systematic prediction errors that are at least partly driven by unforeseen adaptation.