Climate change will lead to massive conflicts, according to claims of such prominent sources as Sir Nicholas Stern and the US National Security Agency - claims repeated by the media. Efforts to tease a specific climate change signal from historical records of civil conflict have proved inconclusive, however: they postulate that farmers will become fighters when resources become critically scarce; but they have been unable to illuminate what specific mechanisms may be involved. Yet the potential for climate change to cause significant civil conflict seems intuitively obvious, and the need for better understanding remains urgent. My research focuses on sub-Saharan Africa, the most conflict-prone region in the world; and it asks what factors make some countries erupt in civil conflict, while others do not. I find that drops in agricultural exports diminish government capacity as tax revenues shrink, leading to an increase in the risk of civil conflict. Thus, government capacity to provide security and services is likely to become weak just at the time when climate change is increasing the need for both. How governments respond will determine the risk of civil conflict, but this research shows that their capacity to respond will, in fact, also be affected. The implications of these conclusions apply beyond sub-Saharan Africa, and begin to move the debate from questions around if climate change will cause conflict to more productive discussions of how climate change may affect conflict risk.