The objective of this study was to predict the amount and the distribution of compression wood (CW) within a Norway spruce [Picea abies (L.) Karst.] plank based on green plank curvature. The findings indicated a possibility of predicting the longitudinal distribution of CW from the green plank curvature. Areas free from CW showed a typical concave shape in relation to the centre of the log, while CW was present when a convex shape was shown. The larger the magnitude of convex curvature, the higher the concentrations of CW that could be found, and a larger fraction of dried planks was rejected due to excessive warp. This study also determined what information can be used to eliminate areas of high concentrations of CW by cutting and how cutting affects the grading results with respect to warp. Over 50% of the plank length showing a high concentration of CW (> 30% of the cross-cut volume) was successfully cut off. Cutting strategies based on predicted CW concentrations resulted in a 10-40% increase in accepted plank length.