Recent literature has shown that crosscutting social cleavages reduce the likelihood of civil war. This article argues that the same logic does not apply to lower-scale group violence such as riots, which differ in such a way that crosscutting social cleavages should often have the opposite effect, increasing both the frequency and scale of riots. We test this argument by analysing Muslim–Christian violence in the post-Suharto era, combining a new subnational data set of ethno-income and ethnogeographic crosscuttingness with a new and comprehensive subnational data set of violence in Indonesia. Our findings suggest that high ethno-income crosscuttingness, when combined with a high degree of urban anonymity and close living quarters, is a potent setting for inter-group communal violence. We conclude with a discussion of how context matters in understanding the effect of macrostructural variables such as crosscuttingness on violence.