The derivative diffuse reflectance UV-vis-NIR spectroscopy combined with the multivariate methods are utilized for the discrimination and classification of the soil samples collected from the north-western part of India. The acquired spectra reveal the presence of different organic and inorganic minerals such as humic acid, fulvic acid, hematite, etc. in varying amounts. The differentiation/segregation among soil samples is achieved by peak comparison and chemometric methods like clustering algorithm and principal component analysis (PCA). Among these, the PCA method gives clear discrimination of soil samples. The developed PCA model is further validated by analyzing unknown samples for the prediction to their respective clusters significantly. Principal component linear discriminant analysis (PC-LDA) based discriminant model is developed to classify the unknown soil samples to its respective groups. PC-LDA based model reveals 95 % accurate clustering of the soil by the leave-one-out cross-validation approach.