This study investigates whether news coverage about unemployment affects people's perceptions of the state of the economy. I compile a German state-level data set, based on household surveys and information obtained from analyzing 35 newspapers. The data are used to separate media effects from real economic consequences, taking advantage of two sources of exogenous variation. First, I exploit the salience of “milestones” in the number of unemployed. The news value of these milestones, which is not based on economic fundamentals, causes the media to report more about unemployment than usually. Second, I show that the amount of reports decreases when competing newsworthy events occur at the time of the release of the monthly unemployment statistics. Instrumental variable estimates indicate that a one standard deviation increase in coverage accounts for about a quarter of the average monthly change in the index of economic perceptions. •Milestones in the number of unemployed increase the amount of unemployment news.•Disasters and terrorist attacks crowd out unemployment reports.•I use an IV approach to estimate the effect on people's perceptions of the economy.•A 1 SD increase in news accounts for a quarter of the monthly change in perceptions.