Density, modulus of elasticity (MOE), and bending strength (MOR) are important properties of structural timber, and knowledge about the variability of these properties is important to make efficient use of the timber. To utilize such information in the production of structural timber, the information must be available before sawing. This study presents models describing the variability of density and bending properties of Norway spruce (Picea abies L. Karst) boards within individual trees, as well as among trees and stands, based on geographical data and forest inventory data including external tree measurements. The models were based on 1551 boards from 17 sites in Southern Norway, Eastern Norway, and Trøndelag. Important variables describing variation in density, MOE, and MOR between sites were site index and elevation. For density, latitude gave additional information. Age, diameter at breast height, and longitudinal position within the tree were the most important variables at the tree level. The models explained major parts of the site variance of all properties, and for MOR, they explained a substantial part of the variance due to trees. In addition to being used for predicting the properties of structural timber from current forest resources, the models also provide information that can be used to predict the effects of silviculture on timber properties in future forest stands.