Machine strength grading of structural timber is based upon relationships between so called indicating properties (IPs) and bending strength. However, such relationships applied on the market today are rather poor. In this paper, new IPs and a new grading method resulting in more precise strength predictions are presented. The local fibre orientation on face and edge surfaces of wooden boards was identified using high resolution laser scanning. In combination with knowledge regarding basic wood material properties for each investigated board, the grain angle information enabled a calculation of the variation of the local MOE in the longitudinal direction of the boards. By integration over cross-sections along the board, an edgewise bending stiffness profile and a longitudinal stiffness profile, respectively, were calculated. A new IP was defined as the lowest bending stiffness determined along the board. For a sample of 105 boards of Norway spruce of dimension 45 × 145 × 3,600 mmÂ3;, a coefficient of determination as high as 0.68-0.71 was achieved between this new IP and bending strength. For the same sample, the coefficient of determination between global MOE, based on the first longitudinal resonance frequency and the board density, and strength was only 0.59. Furthermore, it is shown that improved accuracy when determining the stiffness profiles of boards will lead to even better predictions of bending strength. The results thus motivate both an industrial implementation of the suggested method and further research aiming at more accurately determined board stiffness profiles.