Existing research by experimental psychologists, qualitative sociologists and economics theorists agree on the importance of intergroup bias both in access to inputs and in observed outcomes to explain out-group behaviour, group mobilisation and, ultimately, social unrest. However, between-group or horizontal inequalities (HIs) have received little attention in the lively empirical literature on civil conflict. To an extent, this is not surprising given the daunting task of measuring distances between groups whose boundaries are typically based on non-economic dimensions. The aim of the paper is twofold: to situate HIs within the existing strands of literature on inequality, ethnic diversity and conflict and to investigate empirically the link, if any, between HIs and ethno-communal violence. Using district-level data from Indonesia, the paper finds that horizontal inequality in child mortality rates and its change over time are positively associated with the occurrence of deadly ethno-communal violence, after controlling for a number of intervening factors including population size, ethnic diversity and economic development. The estimated effects are found to be robust to a number of sensitivity tests. Results also suggest that violent conflict is more likely to occur in areas with relatively low levels of economic development and greater religious polarisation. Standard measures of (vertical) income inequality as well as other purely demographic indicators of ethnic diversity are found to have no significant impact on the likelihood of communal violence.