Current research on resource and environmental conflict – and indeed most research in any field within international relations – rarely considers the concept of scale. Instead, empirical analysis is conducted at a convenient, predefined level with little sensitivity to the underlying assumptions of the theoretical propositions. Regardless of the nature of conflict and the qualities of the resources under study, the dominant scale of investigation is the state, including pairs of states (i.e., ‘dyads’). Popular theories about why and where war occurs, however, often refer to local factors and micro-level mechanisms. According to Collier and Hoeffler, rebellion is more likely when opportunity costs are unusually low, such as when valuable resources are available for extortion, when unemployment rates are high, and when the environment is favorable to guerrilla warfare (rough terrain, low population density). Similarly, Fearon and Laitin argue that outbreak of civil war is explained primarily by conditions that favour insurgency (again, poverty and rural areas are identified, but also aspects of ‘state weakness’), whereas Hauge and Ellingsen point to environmental degradation as a significant risk factor. The intensity of such variables may vary considerably across space within the boundaries of states; yet, conflict analyses prevail at the country level. The result is an unfortunate incompatibility in scales between theory and analysis, and this is particularly evident among quantitative studies linking environmental factors with civil war. This essay therefore speaks primarily to that particular literature.Also an output of the Geographic Representation of War Network (GROW-net) project.