We study how individual unemployment expectations are shaped and updated using a unique longitudinal survey data set with subjective unemployment expectations. The survey data is linked with third-party reported administrative data on unemployment realizations, such that we are able to examine how prediction errors lead individuals to update their unemployment expectations. We find that people are constantly uncertain about their unemployment prospects. The uncertainty causes them to adjust their unemployment expectations when their predictions turned out to be incorrect. As a result, people’s expectations concerning future unemployment are not constant and they are heterogeneous across the population at any given point in time. We document that unemployment expectations and prediction errors are important determinants of economic decisions, such as how much to save or whether to insure against earnings losses. Subjective unemployment expectations can thus help explain why people, who are observationally similar, make differently economic decisions.