Large tree individuals in tropical forests store huge amounts of carbon and are thus essential players within the carbon cycle of tropical forests. While many studies have focused on biomass accumulation of tropical forests and their potential to act as a carbon sink, we lack a cohesive study on how different tree sizes contribute to the overall forest carbon flux budget. Here, we use xylem sap flux derived gross primary productivity (GPP) that was modeled with eddy covariance data for an old-growth moist lowland forest in Central Amazon. Additionally, net primary productivity (NPP) was calculated from forest inventory data. We found that GPP was 28.46 Mg C ha-2 yr-1 at our study site. Emergent canopy trees (diameter >30 cm; average height of 28 m) were responsible for 21.47 Mg C ha-2 yr-1 of the overall GPP, whereas subcanopy and understory trees contributed 3.95 Mg C ha-2 yr-1 and 3.04 Mg C ha-2 yr-1, respectively. While emergent canopy trees were characterized by low carbon use efficiency (CUE) for only allocating 23% of their photosynthetic products towards growth, these trees were still responsible for more than half of the overall NPP. Subcanopy and understory trees were more efficient by allocating 67% and 59% of their carbon assimilates towards biomass growth, respectively. GPP showed seasonal patterns with a peak during the dry season. GPP of emergent canopy trees doubled in the dry season, whereas understory and subcanopy trees had an almost 60% higher GPP than the wet season. We found evidence that productivity was more limited by the low evaporative demand than by light. This study provides further evidence for the importance of large trees in tropical moist forests and highlights their crucial role in forest carbon cycling. Due to their high drought-related mortality, large trees will make up a critical component of the response of tropical forests to climate change.