This paper studies empirically the impact of mining on conflicts in Africa. Using novel data, we combine geo-referenced information over the 1997-2010 period on the location and characteristics of violent events and mining extraction of 27 minerals. Working with a grid covering all African countries at a spatial resolution of 0.5x0.5 degree, we find a sizeable impact of mining activity on the probability/intensity of conflict at the local level. This is both true for low-level violence (riots, protests), as well as for organized violence (battles). Our main identification strategy exploits exogenous variations in the minerals' world prices; however the results are robust to various alternative strategies, both in the cross-section and panel dimensions. Our estimates suggest that the historical rise in mineral prices observed over the period has contributed to up to 21 percent of the average country-level violence in Africa. The second part of the paper investigates whether minerals, by increasing the financial capacities of fighting groups, contribute to diffuse violence over time and space, therefore affecting the intensity and duration of wars. We find direct evidence that the appropriation of a mining area by a group increases the probability that this group perpetrates future violence elsewhere. This is consistent with "feasibility" theories of conflict. We also find that secessionist insurgencies are more likely in mining areas, which is in line with recent theories of secessionist conflict.