The empirical evidence from studies linking geographic factors like terrain and natural resources to civil war is generally weak and not robust to varying samples or coding procedures. We argue that these investigations suffer from a major weakness: although most civil wars are geographically limited to small parts of the host countries, the analyses rely almost exclusively on country-level data. We demonstrate how Geographical Information Systems (GIS) can be used to generate precise measures of space-varying factors at the scale of the conflict. A comparison of several relevant variables measured both at the scale of the country and the conflict demonstrates that country statistics are poor approximations of the conflict zones. An analysis of duration of civil war further shows that certain findings are indeed dependent upon the scale of measurement. We conclude by discussing how GIS and spatial analysis may be applied in future research to increase our understanding of location, duration, and risk of armed civil conflict.