Analysis of variance of chromatographic data is usually performed on the peak table or on entire chromatograms. These two data forms require signal pretreatment. Peak table requires peak detection, their standards and quantification, and the second form of data organization requires warping of the studied chromatograms to eliminate the observed peak shifts, which occurs due to minor variations in chromatographic conditions. In our study, a new form of data representation well suited for chromatographic data originating from multi-channel detection is proposed. It requires neither warping of chromatograms, nor peak detection. Its principles and performance are demonstrated for a real data set (being a part of a larger research project initiated to characterize the infusion of fermented rooibos herbal tea in terms of phenolic composition and antioxidant activity). As the method of choice for the analysis of data variation, the Multiple Analysis of Variance applied to the pairwise data representation was chosen. © 2013 Elsevier B.V. All rights reserved.