& Context The commercial feasibility of sawmilling depends on the expected volume and value of sawn planks. Models that predict the volume of sawn timber of a particular quality and produced from logs of known characteristics are therefore very useful. & Aims The objectives were to study variation in sawing yield and to obtain models that predict lumber volume and grade recovery on the basis of easy-to-measure predictor variables of saw logs. & Methods Forty-six oak trees growing in Galicia (NW Spain) were felled and cut into logs. The logs were visually graded and sawn mainly into quartersawn planks, which were dried, planed and visually graded for structural purposes. & Results The total volumetric sawing yield was 47.6 %. The sawing yield for planks of structural dimensions (cross-section, 70×120 or 70×170 mm) was 43.4 %, but decreased to 8.4 % for structural sized and quality grade beams because of wane and biotic damage in many pieces. Log grade did not significantly affect sawing yield in the sample analysed, despite the wide range of diameter over bark at the smallest end in the sampled logs (22-77 cm). The sawing pattern affected total sawing yield (F=4.913; p value=0.001) and the sawing yield for structural planks (F=6.142; p value=0.0002); radial sawing with one cut and live sawing of half logs provided the highest yields. Three models were proposed for estimating sawn volume in timber products, with the small-end log diameter over bark as the predictor variable and R adj 2 between 0.31 and 0.78 (p value<0.01). & Conclusion For the purpose of producing oak timber destined for structural use, the presence of bark and sapwood in planks must be reduced in the sawing process; this would decrease the total lumber recovery but increase the timber value yield. Air drying must be accelerated to reduce biotic damage in sawn planks. Geometric mean diameter over bark at the smallest end (d) outperforms other measures as a predictor variable for total or structural sawn timber volume.