A new empirical physical model is presented to estimate the soil-water characteristic curve approximately. The capillary theory is used to infer capillary water height in granular materials of a specific grain diameter. For most soils, the correction factor introduced in this model shows remarkable linear relationship with the suction water head in logarithmic coordinates and their relationship can be expressed as . Data of 406 soil samples from the unsaturated soil database UNSODA are used to analyze the fitting coefficients and . It is shown that they are affected by neither grain-size distribution nor void ratio, and they might be related to the mineral composition of soils. Predictions of the soil-water characteristic curve (SWCC) of a specific soil can be achieved using the grain-size distribution of the soil, and the parameters and can be calculated from some experimental results of soils with similar mineral compositions. The mean values of a and b calculated in this paper may also be an alternative in the absence of related data.