Quality sorting of sawlogs based on three-dimensional (3D) or X-ray scanning or a multivariate combination of variables from both methods may be used to decrease the production of off-grade products carrying unwanted combinations of dimension and grade. There is, however, potential for further improving the sorting accuracy if 3D and X-ray raw data are combined at an early stage using path length compensation. From the measured 3D shape, a good estimate of the length of each X-ray path through the log can be made, enabling the calculation of a log density profile from the measured X-ray attenuation. The effect of this technique on heartwood diameter measurements of 423 Scots pine (Pinus sylvestris L.) logs was evaluated. By the addition of 3D data to the X-ray data it was possible to raise the predictability of the heartwood diameter from R 2=0.84 to 0.95 and to improve the root mean square error from 17 mm to 9.3 mm, primarily because of the enhanced contrast between heartwood and sapwood.