Tropical dry forests harbor major carbon stocks but are rapidly disappearing due to agricultural expansion and forest degradation. Yet, robustly mapping carbon stocks in tropical dry forests remains challenging due to the structural complexity of these systems on one hand, and the lack of ground data on the other. Here we combine data from optical (MODIS) and radar (Sentinel 1) time series, along with Lidar-based (GEDI) canopy height information, in a Gradient Boosting Regression framework to map aboveground biomass (AGB) in tropical dry forests. We apply this approach across the entire dry Chaco ecoregion (800,000 km2) for the year 2019, using an extensive ground dataset of forest inventory plots for training and independent validation. We then compare our AGB models to structural vegetation parameter such as percent tree and shrub cover, as well as Level-2 data from GEDI. Our best AGB model considered MODIS and Sentinel 1 data, whereas the additional use of GEDI-based canopy height data did not contribute substantially to model performance. The resulting map, the first high-resolution AGB map covering the entire ecoregion, revealed that there are still 4.65 Gt (+/- 0.9 Gt) of AGB in the remaining natural woody vegetation of the Chaco. Nearly three quarter of the remaining AGB in natural vegetation is located outside protected areas, and nearly half of the remaining AGB occurs on land utilized by traditional communities, suggesting considerable co-benefits between protecting traditional livelihoods and carbon stocks. Our models also had a much higher level of agreement with independent ground-data than global AGB products, which translates into a huge, up to 14-fold, underestimation of AGB in the Chaco by global maps in comparison to our regional product. Our map represents the most accurate and fine-scale map for this global deforestation hotspot and reveals substantial risk of continued high carbon emissions should agricultural expansion progress. In addition, by combining our AGB map with structural vegetation parameters we provide for the first time for tropical dry forests an understanding of carbon stocks in relation to the vegetation structure in these ecoregions. More broadly, our analyses reveal the considerable potential of combining time series of optical and radar data for a more reliable mapping of above-ground biomass in tropical dry forests and savannas.