The measurement of above-ground biomass is important to understand carbon flow between trees and the atmosphere; remote sensing plays an important role in making this possible for extensive and hard-to-reach areas. This study compared above-ground forest biomass estimation models using data from different sources, including Landsat ETM+, Aster GDEM, ALS (LiDAR) and forest inventories. Two sets of predictors were established: the first included variables extracted from Landsat ETM+ and Aster GDEM, while the second included variables from Landsat in combination with LiDAR products (Digital Terrain Model, Digital Surface Model and Canopy Height Model). The Random Forest algorithm was used to build all models; this method explicitly returns the importance of each predictor and therefore allows the selection of the best set of variables. Estimations were made separately by forest cover for Pinus radiata, Eucalyptus globulus and second-growth Nothofagus glauca. Better results were obtained using the combination Landsat-LiDAR than those using Landsat-Aster GDEM data. Also, the results were better when applying the model to pine cover (pseudo R2 77.22%).