Spectroscopic fingerprinting ( 1 H/ 13 C NMR, mid-infrared FT-IR, and near-infrared NIR) combined with multivariate chemometrics has matured into a core analytical paradigm for rapid screening, classification, and quantification of fraudulent and adulterated matrices across the agro-food chain. This review synthesizes advances (2005–2025) in instrumentation, spectral preprocessing, algorithmic modeling, validation strategies and data fusion, with particular emphasis on robustness, transferability and on-site deployment. We discuss matrix-specific challenges (oils, dairy, beverages, seeds, feeds, fertilizers), summarize performance metrics reported in the literature, and identify critical gaps (model generalizability, open-world detection, standardized reference libraries) that constrain regulatory uptake. Recent comparative and systematic reviews and domain-specific studies are used to underpin recommendations for best practice. • Spectroscopic fingerprinting enables rapid, non-targeted detection of food adulteration. • NMR, FT-IR and NIR provide complementary chemical information for authenticity control. • Chemometric workflows enhance classification accuracy and adulterant quantification. • Calibration transfer and external validation remain critical for regulatory deployment. • Data fusion and deep learning improve sensitivity for complex or concealed frauds. • Standardized protocols and shared spectral libraries are essential for reproducibility.
Vicente et al. (Sun,) studied this question.