Models are abstractions that enable to assess and predict forest stands variables. Two broad methods to estimate biomass were defined. The direct method, the most accurate, has the disadvantage of resulting from destructive sampling. Inversely, the indirect method uses a variety of mathematical methods, with forest inventory, remote sensing, and ancillary data as explanatory variables. The accuracy of the biomass models is dependent on data acquisition precision and accuracy as well as on the model’s uncertainties. Moreover, model accuracy is also dependent on species, individual tree biomass partitioning, stand structure, region, and spatial and temporal scales. This chapter overviews the data sets and mathematical methods used for modelling biomass and their uncertainties. Overall, the performance of the forest biomass functions is linked to its ability to accommodate the variability inherent to forest data and to make biomass assessments, monitoring, and predictions with the best possible precision and accuracy and the smallest bias.