Inulin, protein, total polyphenols, and ash are important physicochemical components of the Jerusalem artichoke. They have an impact on Jerusalem artichoke’s processing and nutritional characteristics. Traditional detection methods for these components often require a large number of chemical reagents and high-standard equipment. To achieve rapid prediction of the main physicochemical indicators in the Jerusalem artichoke, 56 Jerusalem artichoke samples were examined in this study. Spectral data were collected in the range of 900~2 500 nm by using a near-infrared (NIR) spectrometer, and partial least squares regression prediction models were established in combination with the measured contents of inulin, protein, total polyphenols, and ash. The results showed that, after outlier elimination, spectral preprocessing, and characteristic band screening, prediction correlation coefficients (R2p) for the optimal quantitative models of inulin, protein, total polyphenols, and ash were 0.869 5, 0.930 7, 0.943 3, and 0.920 6, respectively. The corresponding root mean square errors of prediction were 2.237 2 g/100 g, 0.363 4 g/100 g, 0.107 7 mg.g-1, and 0.158 1 g/100 g, respectively, indicating good predictive ability of the models. External validation showed no significant difference (P>0.05) between the detection results obtained using NIR spectroscopy and standard chemical methods. This study provides a theoretical basis for the application of NIR in the rapid prediction of physicochemical components in the Jerusalem artichoke.