We study whether salient media coverage of refugees drowning in the Mediterranean affects individual xenophobic attitudes. We combine a randomized survey experiment - a variant of the classic trolley dilemma' - that implicitly elicits individual attitudes towards foreigners, with variation in interview timing, and find that such issue salience significantly decreases xenophobic attitudes by 2.2 percentage points. Our results thus support the idea that exposure to news describing immigrants as victims (instead of a threat) can significantly affect public opinion and mitigate bias against immigrants.