Although conflict incidence is likely to be characterized by spatial dependence, the scientific literature on conflicts often neglects the issue thus, implicitly, assuming independence among observations. We argue that such assumption could lead to biased and inconsistent results and we provide an exemplary application to the case of the Mano River Region (MRR) in West Africa. Once we detected the existence of spatial dependence within the distribution of conflict incidence, we introduce spatial econometrics techniques in order to explore diffusion paths of violence within the region. We firstly project on a spatially disaggregated map, built as a regular grid, the conflict occurrence and several georeferenced determinants of civil conflicts. Then, we model spatial dependence through the introduction of spatial autoregressive terms on both dependent and independent variables (SAR and SD Models). Across several models, civil conflict is found steadily clustered in space with significant spill-over effects on contiguous locations. Among other determinants, natural resources – namely diamonds and gold – are confirmed as relevant drivers of conflict diffusion and show neighbouring effects since their location and proximity may affect conflict dynamics.