Misaligned mediation leverage is developed as a framework for understanding mediation failure in protracted conflicts, focusing on leverage, consent, and the political economy of negotiation spoilers. The argument advanced is that mediation fails when leverage is directed at formal principals while the capacity to violate agreements is dispersed across semi-autonomous commanders whose incentives diverge from those of negotiators at the table. Drawing on process tracing of IGAD-led mediation processes (2014–2018); comparative analysis of Angola’s Lusaka Protocol (1994) and Liberia’s Accra Agreement (2003); mediation team interviews; and application of formal bargaining models under incomplete information, the study engages debates in negotiation theory, spoiler theory, and principal–agent problems in international mediation. It develops a model of how mediator leverage interacts with internal factional dynamics to explain why mediated agreements collapse. Three core claims are advanced. First, the 2015 ARCSS failed not because mediation was absent, but because leverage and consent were concentrated at the leadership level while field commanders retained both the incentives and capacity to defect. Second, spoiler behaviour is better explained by principal–agent divergence within armed movements than by a simple binary between peace-seekers and rejectionists. Third, sanctions and coercive diplomacy are effective only when paired with enforcement arrangements that reshape the calculations of commanders, financiers, and external patrons simultaneously. The study concludes that peace processes must map factional chains of control rather than assume that elite signatures ensure compliance. It further emphasizes that effective leverage must extend beyond top leadership to reach