A novel method of three-dimensionally visualizing the internal structure of a soybean seed, which was based on excitation-emission matrix (EEM) measurement and EEM pattern analysis, was developed. Cross-sectional images of three soybean samples at different germination stages were taken at 561 different excitation and emission wavelength conditions so that the EEM at each pixel was calculated. By applying principal component analysis (PCA) to these EEM data, two wavelength condition sets were selected for a simplified EEM measurement. Serial cross-sectional images at various depths of soybean seeds at pre-, early, and late germination stages were taken under the two wavelength condition sets. The difference in the simplified EEM pattern was converted into the difference in color by PCA and the color assignment of each pixel. Then, cross-sectional images were accumulated by a volume rendering method to develop three-dimensionally reconstructed models. Finally, six different three-dimensional reconstructed models of samples at each germination stage under the two wavelength condition sets were developed. Observations of these images suggested that vascular bundles or constituents forming them already existed in soybean seeds at the pre-germination stage. It was also considered that a soybean seed, similar to a cereal seed, has a specific part that emits autofluorescence between the seed coat and the endosperm. In addition, it was supposed that, at the early germination stage of a soybean seed, chlorophyll or related constituents were expressed in the cotyledon on the hypocotyl side, not throughout the seed, and spread along the vascular-bundle-like structure as germination progressed.