Abstract: This article focuses on the foundation and validation of soil salinity model based on electro magnetic (EM) induction technique. Variance contribution of factors (such as soil salt content, soil water content, soil density, clay content, electrical conductivity of groundwater, and groundwater depth) to soil apparent electrical conductivity (eq. EM value) measured by electromagnetic inducting instruments and the order of function weights of influence factors were obtained by path analysis. According to variance contribution, their relative weight order from high to low and designed total contribution (90%), dominant non-salinity factors to EM values were selected and thus parameter system of the model for interpreting soil salinity were built up. Optimized layer-based (0-20, ＞20-60, ＞60-100 and ＞100-160 cm) soil salinity model was found via stepwise regression of multi-factors and interactions. Validation result showed that the model was of high accuracy and almost all of the relative errors were lower than 10%. This model can be used to interpret soil salinity in inner or coastal irrigation areas with shallow groundwater table.