【Objective】This study aimed to provide a scientific basis for rapid monitoring and regionalized classification-based reclamation of salinized soils by constructing an inversion model for electrical conductivity (EC1:5) of soil profile extracts in salinized areas of Northwest China.【Method】Typical saline-alkali farmland in the Hetao Irrigation District was selected as the study area. Electrical resistivity tomography (ERT) was used to obtain soil resistivity profiles with lengths of 35.5~71 m and depths of 0–6 m. Meanwhile, nine boreholes (3~5 m deep) were sampled by layers to measure soil EC1:5, water content, and bulk density. Based on pedogenetic theory, soil resistivity, depth, water content, and bulk density were selected as input variables. Random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost), and linear regression (LR) algorithms were applied to perform EC1:5 inversion at a field scale.【Result】The results indicate that soil resistivity and depth were the dominant variables controlling EC1:5 inversion, while water content and bulk density had relatively minor contributions. Among the models, the RF model achieved the best performance, with an independent validation R² of 0.65 and a root mean square error (RMSE) of 56.51 μS·cm⁻¹, effectively matching the vertical variation characteristics of electrical conductivity measured in the nine boreholes. The inverted soil EC1:5 exhibited significant spatial heterogeneity on the two-dimensional profile, ranging from 170.58 to 441.01 μS·cm⁻¹, with a mean value of 275.48 ± 58.81 μS·cm⁻¹, which is opposite to the variation characteristics of soil resistivity.【Conclusion】The ERT-based inversion method for soil profile EC1:5 demonstrated good applicability and can effectively compensate for the limited vertical resolution of electromagnetic induction (EMI) techniques. It provides reliable technology for the rapid acquisition of high-resolution salinity information in soil profiles.