How does conflict affect firms’ investment decisions? Past results are mixed: a third of studies we reviewed report null or mixed correlations; some suggest conflict increases investment. We rationalize these results, arguing that armed conflict has divergent effects depending on firms’ exposure to violence. Conflict can deter investment by disrupting production or raising uncertainty. Yet, conflict can encourage investment by hampering government oversight. We argue each mechanism operates over different geographic extents. We use data from the mining sector to test these claims and report three main results. Firms operating at conflict sites dramatically reduce investments. By contrast, firms operating in territory surrounding conflict, but at a remove from fighting, actually increase investment. Firms far from violence see a small negative effect. These divergent responses cannot be inferred from aggregate flows: we show conflict depresses aggregate investment, but this reflects responses among firms far from fighting.