<p>China’s innovative Farmland Use Regulation (FUR) system, implemented since 2020, represents a critical effort to curb long-term farmland decline. However, existing literature has yet to empirically evaluate the effectiveness of this complex policy mix and its underlying governance mechanisms. To bridge this gap, this study argues that the effectiveness of the FUR is not uniform but is fundamentally moderated by the underlying Principal-Agent (P-A) dynamics, where policy impacts are contingent on the alignment of incentives between central and local actors. Using a panel dataset from 286 Chinese cities (2019-2022), we employ a two-way fixed-effects model, complemented by instrumental variable strategies and Geo Detector analysis, to assess the policy’s effectiveness. Results suggest that the FUR framework has significantly reduced net farmland loss. More importantly, its success is driven by mechanisms that resolve P-A conflicts by increasing oversight (technological monitoring) and raising non-compliance costs (recultivation enforcement). Conversely, mechanisms reliant on local personnel, where goal conflicts are more acute, prove less effective. Heterogeneity analysis further reveals that the policy is most potent in major agricultural regions where the principal’s and agents’ objectives are more naturally aligned. These findings deepen the understanding of how top-down regulations perform in multi-level governance systems, helping researchers and policymakers anticipate the spatial variance and effectiveness of similar environmental policies.</p>