Red fleshed apples (Malus × domestica Borkh.) differ in colour intensity between cultivars, seasons and sites. The objective of this study was to develop a procedure for predicting anthocyanin content from digital images of flesh discs. Flesh cylinders of uniform colour were excised, scanned and their colours determined in the R, G and B and the L*a*b* colour spaces. Anthocyanin content was also quantified chemically. A calibration line was constructed to predict anthocyanin content of flesh discs of varying colour from a scan or a photograph in the studio or outdoors. Anthocyanin concentration was linearly related to the logarithms of G, B and L*. From these relationships, the anthocyanin content of a flesh disc was predicted, pixel by pixel. Colour corrections were applied using a reference colour chart included in all images. The Finlayson algorithm was most effective for correcting the G parameter obtained by a flatbed scanner. For variable imaging methods (scanning or photography), the Vandermonde algorithm for correcting the L* parameter and the Finlayson algorithm for correcting the G parameter were most effective in predicting anthocyanin content. The procedure allows accurate prediction of anthocyanin content of red fleshed apples from simple colour scans or photographs.