Abstract: Spillovers of deforestation activities to untargeted actors and regions have the potential to greatly reduce the effectiveness of zero-deforestation supply chain commitments (ZDCs). Likewise, such spillovers create the potential of augmenting livelihood outcomes for marginalized groups. While understanding of the direct impacts of supply chain policies has increased, the degree to which FLARE FLARE 2022 ROMEdeforestation “leakage” occurs remains unclear due to methodological challenges and limited data availability. Focusing on the beef cattle sector, the largest driver of tropical deforestation globally, we use newly assembled temporally and spatially explicit property-level data on cattle sales and deforestation for the Brazilian state of Pará to better understand the processes leading to low ZDC effectiveness and local leakage, as well as the potential livelihood affects this might have. We find that incomplete adoption of ZDCs among cattle buyers allows producers to avoid ZDC policies and continue deforesting, accounting for 74% (450,273 ha) of the deforestation detected in our study. Yet laundering, whereby indirect suppliers to ZDC companies to whom ZDCs are not yet implemented continue to deforest and sell through “clean” direct suppliers is also linked to 96,311 ha of deforestation. This laundering appears to drive policy leakage, as direct suppliers of ZDC companies are significantly more likely to switch to an indirect ZDC supplier role after deforesting than direct ZDC suppliers who do not deforest. We find that these indirect suppliers linked to leakage processes are more likely to be small, more marginal producers, far from urban centers. These results suggest that enforcing ZDC requirements among indirect suppliers is critical to meet the direct goals of supply chain policies, yet will likely have negative livelihood impacts. Therefore, measures that seek to close leakage pathways and increase ZDC effectiveness should also include inclusive measures for the more marginal producers that would be disproportionately affected by these policies.