【Objective】Hyperspectral remote sensing technology holds considerable research value for monitoring and assessing heavy metal contamination in soils. However, it is unclear how this technology can be used to detect different heavy metal fractions in soil. 【Method】This study collected 22 intact soil profile samples with depths of approximately 100 cm from farmland soils in the Le""an River Basin, Jiangxi Province, China. The samples were used to investigate the potential of hyperspectral imaging (HSI, 400-1010 nm) for predicting five copper (Cu) fractions in intact soil profiles, including total Cu, weak acid-extractable Cu (F1), reducible Cu (F2), oxidizable Cu (F3), and residual Cu (F4). After the spectral data and Cu contents of the soil profile samples were measured, prediction models for soil Cu contents were established. Several modeling methods were applied to investigate the effect of different spectral preprocessing techniques on prediction accuracy, including partial least squares regression (PLSR), random forest (RF), Cubist regression tree (Cubist), Gaussian process regression (GPR), and Support vector machine (SVM). 【Result】The results show that the four machine learning algorithms, namely RF, Cubist, GPR, and SVM, generally outperformed the linear PLSR model in terms of R², demonstrating higher predictive accuracy. After preprocessing with the combined absorbance transformation and first derivative method (Abs+FD), the SVM-based model achieved relatively good predictive performance for the five soil Cu fractions in the independent validation set (F1: R2p = 0.78, RMSEp = 0.56 mg·kg-1; F2: R2p = 0.78, RMSEp = 0.40 mg·kg-1; F3: R2p = 0.67, RMSEp = 1.33 mg·kg-1; F4: R2p = 0.70, RMSEp = 2.91 mg·kg-1; Total Cu: R2p = 0.67, RMSEp = 3.64 mg·kg-1). 【Conclusion】These findings indicate that HIS combined with machine learning can effectively predict multiple heavy metal fractions in soil profiles, which is of great significance for improving our understanding of the migration and transformation of heavy metals in soil and for conducting regional soil pollution risk assessments.