Elevated groundwater salinity associated with produced water, leaching from landfills or secondary salinity can degrade arable soils and potable water resources. Direct‑push electrical conductivity (EC) profiling enables rapid, relatively inexpensive, high-resolution in-situ measurements of subsurface salinity, without requiring core collection or installation of groundwater wells. However, because the direct-push tool measures the bulk EC of both solid and liquid phases (ECa), incorporation of ECa data into regional or historical groundwater data sets requires the prediction of pore water EC (ECw) or chloride (Cl-) concentrations from measured ECa. Statistical linear regression and physically-based models (PBMs) for predicting ECw and Cl- from ECa profiles were tested on a brine plume in central Saskatchewan, Canada. The PBMs tested included two distinct mathematical approaches for estimating ECw from ECa; a linear model developed in soil sciences, and a power-law model developed in the oil and gas industry (Archie’s Law). A linear relationship between ECa/ECw and porosity was more accurate for predicting ECw and Cl- concentrations than a power-law relationship. Despite clay contents of up to 96 %, the addition of terms to account for electrical conductance in the solid phase did not improve model predictions . In the absence of porosity data, statistical linear regression models adequately predicted ECw and Cl- concentrations from direct-push ECa profiles. These statistical models can be used to predict ECw in the absence of lithologic data and will be particularly useful for initial site assessments. The more accurate linear physically-based model can be used to predict ECw and Cl- as porosity data become available and the site-specific ECw–Cl- relationship is determined.