We use scanner data of supermarket sales to investigate the effects of the EU School Fruit campaign, conducted in a sample of primary schools in the city of Rome during 2010 and 2011, on the consumption of unhealthy snacks. We allocate supermarkets to treatment and control groups depending on whether they are located or not near treated schools and estimate the causal effect of the program by comparing the changes in the sales of snacks in treated stores with the changes in control stores. We find evidence that the campaign reduced the consumption of unhealthy snacks bought in stores located in high income areas. No effect is found in poorer areas. Repeated treatment does not strengthen the effects of the program.