Meta-attitude inaccuracy – the extent to which our perceptions of what the outgroup thinks diverges from what they actually think – fuels intergroup hostility, yet its precursors have been rarely studied. Through four correlational studies, we examined whether and how intergroup contact predicts meta-attitude (in)accuracy in three conflict contexts (Black/White British in the UK, Turks/Kurds in Turkey, and Catholics/Protestants in Northern Ireland). Regardless of context or status, all groups perceived the outgroup to be more negative towards them than in reality. Positive contact predicted increased meta-attitude accuracy (through increased shared reality) which was in turn associated with more positive outgroup attitudes. The role of negative contact and affective mediators (such as intergroup anxiety and fear) was not consistent and depended on group and context. The potential use of positive intergroup contact as a rectifier of overestimated meta-attitudes in conflict contexts is discussed.