The ways in which image classification can be utilised in automatic colour correction are discussed. Before images can be classified, a restricted set of numeric features must be extracted from image data. Many of these features can be defined on the basis of the statistical distribution of the colour values of pixels. In some cases, however, spatial image properties are also needed. In automatic colour correction, image classification guides the selections made within and between different correction elements including the adjustment of primary colour components, adjustment of tone rendering in Lsa colour space, grey balance adjustment and skin colour correction.