It is possible today to detect the internal structure of a log by using X-ray scanning technology. By visual inspection of a longitudinal radiograph image (LRI) of a log, it is possible to manually predict the coming grade of the centre planks. The objective of this study was to identify the features of Scots pine logs visible in the LRI's that were of importance in a manual grading process. The identification of useful features was determined by a survey among respondents connected to the wood and sawmill industry. The test logs originate from the Swedish stem bank and the evaluation of the results was based on the statistical method of partial least square regression (PLS). This study shows that useful indicators of the true grade of the centre planks were the knots and knot-related features as well as the butt swell and the heart wood fraction of the log.