Emotional contagion has recently been described in fish but whether it is affected by familiarity is not known. We tested whether the sight of a distressed conspecific elicited fear in zebrafish, and whether this was modulated by familiarity. Groups of six zebrafish were housed together in the same tanks for 7 days to create familiar conditions. The behaviour of individual fish was then recorded in paired tanks within sight of either a familiar or an unfamiliar individual, before and after distilled water or an alarm substance was added to the demonstrator, but not to the observer. As expected, addition of distilled water did not elicit any behavioural change in either the demonstrator or the observer. However, addition of an alarm cue triggered anti-predatory behaviours in the demonstrator which caused the expression of anti-predatory behaviours in the observer, suggesting the existence of fear contagion. Furthermore, the extent of fear contagion was affected by familiarity, and observers were more active, swam closer to the bottom and further away from the demonstrator when they watched a distressed familiar neighbour than when they watched an unfamiliar fish. Our results have implications for fish welfare because they show that fish can become stressed by simply watching others become stressed. They also have implications for experimental design because fish housed in separate tanks cannot be assumed to be statistically independent if they can eavesdrop on their neighbours.