Abstract: Traditional chemical composition analysis methods for tobacco leaf such as chromatography and mass spectrometry, can accurately determine specific components, but their destructive nature, low throughput and high cost limit their applicability in large-scale and time-sensitive industrial traceability scenarios. Near-infrared spectroscopy (NIRS) technology with its advantages of rapidity, nondestructiveness and low cost can overcome the limitations of traditional analytical methods and has emerged as an ideal tool for tobacco leaf traceability. This review discussed the status and shift in mainstream research models for NIRS-based tobacco origin tracking, from early spectral fingerprint matching to current "spectra→chemical components→origin", driven by large dataset. In this context, the application of this research method in different tobacco raw materials and its core advantages in terms of model robustness and chemical interpretability were discussed. Based on the future research trends, the summary and prospects of this technology in building global databases, deepening chemical information dimensions and integrating with portable devices were discussed.