This paper examines how differences in state's ‘embedded autonomy’ (i.e., the historical relationship between farmers and the state, public institutions, and state capacity) in El Salvador, Honduras, and Nicaragua shaped the management of the 2012 coffee leaf rust (CLR) epidemic, resulting in differential responses. We conducted 80 interviews with key coffee-sector actors and six focus groups representing 20 farmer organizations and 19 technical assistance providers to understand these divergent responses. We first examined the dynamic, historical, and relational contexts of contradictions, contestation, and negotiation in agri-food networks influencing each response. We then explored how the role of the state and agrarian structures in each country affected farmers' performance and access to different support systems for renovating and rehabilitating coffee plantations in response to the CLR epidemic. In El Salvador and Nicaragua, oligarchic control, weak state relationships with small coffee farmers and fragmented bureaucracies hindered effective CLR crisis response and recovery. Although farmers' organizations, NGOs, and the private sector attempt to fill the gap using market or ‘user pay’ mechanisms for CLR management, they cannot fully compensate for the state's shortcomings. In contrast, Honduras, which is closest to a ‘developmental state’ in the coffee sector, performed better. This is attributed to its more pro-smallholder development path since the 1970s, which bolstered state institutional structures and fostered social ties and dynamism within smallholder organizations in the coffee sector. However, decentralization and clientelist relations limited the country's ability to propose long-term CLR management strategies. We argue that a state's ‘embedded autonomy’ is critical in determining its approach and effectiveness in managing CLR crises and collaborating with and supporting coffee farmers. Finally, we highlight the importance of considering both conjunctural factors (such as meteorological and economic conditions) and the state's historical capacity and embeddedness in analyzing CLR outbreaks and their effects across different countries.