Empirical studies on trade and conflict generally make use of highly aggregated data. There are, however, good theoretical arguments to suspect that trade in some goods should have a bigger impact on the likelihood of conflict than trade in others. This article examines the relationship between trade and conflict at a lower level of aggregation. It discusses various theoretical arguments why the relationship between trade and conflict should vary across industry sectors, particularly strategic trade, asset specificity, and possible appropriation of gains from trade. The central hypotheses are tested using dyadic United Nations trade data disaggregated at industry level from 1970 to 1997. The main findings are that trade generally reduces the likelihood of conflict, the relationship is weaker for commodities that are more easily appropriable by force, and the relationship is generally stronger for manufactured goods with the notable exceptions of chemical and metal industries and the high-technology sector.