Soil water repellency (WR) occurs worldwide and affects hydrologic processes such as infiltration, preferential flow, and surface erosion. The degree of WR varies with soil organic C (SOC) and water contents. In this study, we measured WR (by ethanol molarity) as a function of moisture conditions for two soil profiles (17 layers, of which 13 exhibited WR), representing different vegetation and SOC between 0.6 and 14%. Generally, WR was found at SOC ≥2%. Based on measured data, a two-region water repellency (TRWR) model was developed. The model assumes two linear regions in a WR vs. pF (=log[−ψ], where ψ is the soil water matric potential in centimeters of H2O) plot, with linear increase in WR from the moisture content where WR first occurs during drying to the maximum WR at pFWR-max, and a linear decrease from pFWR-max until ambient air-dried conditions. The van Genuchten soil water retention model was used to convert WR–θ (where θ is the volumetric water content) to WR–pF. The TRWR model fitting parameters, slopes, and intercepts, were all highly correlated with SOC (R2 > 0.8). The TRWR model was tested against an independent data set for five soils with 2 to 12% SOC and predicted well the measured WR–θ and WR–pF relations. For high-SOC surface soils, the TRWR model seems promising to predict WR from fully wettable to ambient air-dried conditions, i.e., within the interval where WR-induced fingered water flow probably occurs. Finally, our data imply that clay saturation by SOC (quantified by the so-called Dexter index) is useful for predicting if soils are likely to exhibit WR.