Indiscriminate violence against civilians has long been viewed as a catalyst for new rounds of violence in civil wars. Can humanitarian assistance reduce violence after civilians have been harmed? Crossnational studies are pessimistic, drawing a connection between humanitarian aid and increased civil war violence, lethality, and duration. To date, however, we have few subnational studies of wartime aid and subsequent violence. To examine this relationship, I draw on the Afghan Civilian Assistance Program (ACAP II), a USAID-funded initiative that investigated 1,061 civilian casualty incidents (2011–13). Aid was assigned as if randomly to about half (55.8%) of these incidents, facilitating counterfactual estimation of how assistance affected Taliban attacks against the International Security Assistance Force, Afghan forces, and civilians. Challenging prior studies, I find that ACAP was associated with an average 23 percent reduction in attacks against ISAF, but not Afghan forces or civilians, at the village level for up to two years after the initial incident.