In democracies, voters rely on media outlets to learn about politically salient issues. This raises an important question: how strongly can media affect public perceptions? This paper uses a natural experiment - the staggered introduction of the Digital TV signal in Italy - to measure the effect of media persuasion on the perceptions individuals hold. We focus on crime perceptions and, combining channel-specific viewership and content data, we show that the reduced exposure to channels characterized by high levels of crime reporting decreases individual concerns about crime. The effect is driven by individuals aged 50 and over, who turn out to be more exposed to television while using other sources of information less frequently. Finally, we provide evidence about the effect of the digital introduction on public policies closely related to crime perceptions and on voting behaviour. (C) 2018 Elsevier B.V. All rights reserved.