Characterizing the relationships between the components and taste differences among soy sauces can help evaluate and improve the quality of soy sauces. Although previous studies have reported that certain taste-active dipeptides, the relationships between taste differences and dipeptides of soy sauces are unknown. Therefore, our objective in this study was to investigate the correlations between the dipeptides and the taste differences among soy sauces. To analyze the dipeptides, we constructed an analytical method using liquid chromatography/tandem mass spectrometry (LC/MS/MS) in multiple reaction monitoring mode. Based on this method, we detected 237 dipeptides, the largest number ever detected in soy sauce research. Next, orthogonal projections to latent structures regressions were performed. The data matrix of components, including dipeptides and other low-molecular-weight hydrophilic components obtained from gas chromatography/mass spectrometry (GC/MS), served as explanatory variables (366 in total), whereas a sensory data matrix obtained using quantitative descriptive analysis served as the response variable. The accuracy of models for the sweetness and saltiness differences constructed using the LC/MS/MS and GC/MS data matrix were higher than did models constructed using only the GC/MS data matrix. As a result of investigation of the correlation between the dipeptides and taste differences among soy sauces by using variable importance in the projection (VIP) score, many dipeptides showed the high correlation with taste differences. Specifically, Ile-Gln, Pro-Lys, Ile-Glu, Thr-Phe, and Leu-Gln showed the high VIP score on sweet differences. This study is the first report that reveals the correlations between the dipeptides and taste differences among soy sauces. (C) 2013, The Society for Biotechnology, Japan. All rights reserved.