The aim of this study was to compare the performance of different supervised discrimination methods based on IR data for the classification of starches according to the type of chemical modification undergone. The goal of the supervised classification methods is to develop classification rules. Representative samples of each group (known beforehand) were available, from which the relevant characteristics (chemical modification) were known. On the basis of a training data set, classification rules are determined, which can then be applied to classify new (unknown) samples.