We show that armed conflict in Uganda affects social capital as measured by trust and associational membership. Relying on three rounds of nationally representative individual-level data bracketing a large number of violent events, we find that self-reported generalized trust and associational membership decreased during the conflict in districts in which violent events took place. But we also find evidence for a rapid recovery of social capital in the aftermath of violence. Results from a variety of identification strategies, including difference-in-differences and instrumental variable estimates, suggests that these relationships are causal.