An improved model for estimating nutrient contents of the commercial portion of tree boles was developed for four boreal tree species (Populus tremuloidesMichx.,Betula papyriferaMarsh.,Picea glauca(Moench) Voss, andAbies balsamea(L.) Mill.). This model considers the spatial pattern of variation of nutrient concentrations inside the bole and its relationships with tree size. For all species-nutrient combinations, no significant pattern was found for vertical variations in nutrient concentrations, while two types of nonlinear models, using distance from the tree periphery as the independent variable, fit the pattern of horizontal (or radial) variations. These patterns of variability were used to estimate the global nutrient concentration of the bole by using mathematical integration. The values obtained with this method were generally lower, especially for large stems, than values obtained with traditional methods that do not consider the variability of nutrient concentrations inside the bole. This improved model would permit better estimates of the amounts of nutrients lost in biomass upon forest harvesting, as well as internal cycling of nutrients within the bole.