Commodity agriculture continues to spread into tropical dry forests globally, eroding their social-ecological integrity. Understanding where deforestation frontiers expand, and which impacts this process triggers, is thus important for sustainability planning. We reconstructed past land-system change (1985–2015) and simulated alternative land-system futures (2015–2045) for the Gran Chaco, a 1.1 million km2 global deforestation hotspot with high biological and cultural diversity. We co-developed nine plausible future land-system scenarios, consisting of three contrasting policy narratives (Agribusiness, Ecomodernism, and Integration) and three agricultural expansion rates (high, medium, and low). We assessed the social-ecological impacts of our scenarios by comparing them with current biodiversity, carbon density, and areas used by forest-dependent people. Our analyses revealed four major insights. First, intensified agriculture and mosaics of agriculture and remaining natural vegetation have replaced large swaths of woodland since 1985. Second, simulated land-system futures until 2045 revealed potential hotspots of natural vegetation loss (e.g. western and southern Argentinian Chaco, western Paraguayan Chaco), both due to the continued expansion of existing agricultural frontiers and the emergence of new ones. Third, the strongest social-ecological impacts were consistently connected to the Agribusiness scenarios, while impacts were lower for the Ecomodernism and Integration scenarios. Scenarios based on our Integration narrative led to lower social impacts, while Ecomodernism had lower ecological impacts. Fourth, comparing recent land change with our simulations showed that 10% of the Chaco is on a pathway consistent with our Agribusiness narrative, associated with adverse social-ecological impacts. Our results highlight that much is still at stake in the Chaco. Stricter land-use and conservation planning are urgently needed to avoid adverse social-ecological outcomes, and our results charting the option space of plausible land-system futures can support such planning.