Research on conflict tends to oscillate between two poles: cross-country research using statistical regressions to isolate key variables, and ethnographic research on single countries. I argue that, while the former often fails to capture the political and social dimensions of conflict, the latter can suffer from a lack of causal rigour. I illustrate this argument in the Democratic Republic of the Congo, where I suggest a greater focus on the locus and process of the conflict. Armed groups have become much more integrated into regional political and economic elites over the past two decades, just as the Congolese state and army have themselves developed links with armed groups since the 2003–06 transition. These dynamics, and not a narrow focus on conflict minerals and local violence, should attract greater scrutiny.