Soybean characterization is essential to ensure product quality during distribution according to internal values. In this context, non-destructive characterization method, such as spectroscopy, offer an effective and efficient approach to testing soybean quality in field applications. Among the instruments that are widely used for testing soybean quality, the semi-portable visible near-infrared (Vis-NIR) spectrometer operating at a specific range of 345 to 1033 nm has been proven effective. Therefore, this study aimed to investigate soybean seeds characterization using Vis-NIR spectroscopy with PCA and PLSR chemometric methods. The investigation was carried out using soybean seeds consisting of eight varieties locally produced on Java Island, Indonesia, including Dega1, Dena1, Deja2, Dering1, Devon1, Yellow Flap, Green, and Detam4, in the form of intact, crumble, flour, and paste. Several quality parameters such as protein, fat, crude fiber, carbohydrate, ash, water, chlorophyll, total carotene, vitamin C, and L*, a*, and b* values were measured across intact, crumble, flour, and paste samples. The results of Principal Component Analysis (PCA) showed that sample form and genotypes affected soybean classification. Furthermore, Partial Least Squares Regression (PLSR) showed adequate model calibration for crude fiber, chlorophyll, total carotene, and vitamin C parameters. Based on this analysis, it could be concluded that Vis-NIR spectroscopy proved to be suitable for the classification and prediction of soybean characterization.