The role of managed forests will be essential in addressing the challenges of climate change mitigation by adaptive forest management and enhancement of carbon sinks. The first step is the quantification of the forest carbon budget at large scales, which is an issue at the center of forest landscape ecology. In the aim of estimating regional-scale carbon budgets of Eucalyptus plantations in south-eastern Brazil, the G'Day ecosystem model was combined with remotely-sensed estimates of leaf area index. The spatialization potential of G'Day was assessed through simulations on 16 stands, which encompassed a large range of age and fertility levels. In parallel, the leaf area index (LAI), a key model variable, was obtained for the 16 sites by inversions of MODIS remotely-sensed reflectance time series. These inversions involved the coupling of a hybrid-type canopy radiative transfer model with a soil reflectance model and a leaf reflectance and transmittance model. The inverted LAI was highly variable, both seasonally and interannually. The inverted LAI is used as a forcing variable of G'Day. Results show that the G'Day model is efficient at simulating stand biomass over a wide range of values. Stem biomass increments are also fairly well simulated at different ages, and improved when we use the inverted LAI. A limitation is that inter-stand variability in biomass increment is not well reproduced for the oldest stands. We will discuss the implications of our work for the carbon budget monitoring of Eucalyptus plantations at large scales.