Our research seeks to investigate genomic diversity of landraces of millet, addressing a key uncertainty that will provide a framework for (i) a DNA barcode method that could be used for fast, sensitive, and accurate identification of millet landraces, and (ii) millet landrace conservation including biocultural diversity. We found considerable intraspecific variation among 15 landraces representing six species of small millets using nuclear regions (ITS, ITS1, and ITS2); there was no variation in plastid regions (rbcL, matK, and trnH-psbA). An efficacious ITS2 DNA barcode was used to make 100% accurate landrace assignments for 150 blind samples representing 15 landraces. Our research revealed that genomic variation is aligned with a fine-scale classification of landraces using traditional knowledge (TK) of local farmers. The landrace classification was highly correlated with traits (morphological, agricultural, and cultural utility) associated with considerable factors such as yield, drought tolerance, growing season, medicinal properties, and nutrition. This could provide a DNA-based model for conservation of genetic diversity and the associated bicultural diversity (TK) of millet landraces, which has sustained marginal farming communities in harsh environments for many generations.