Soil hydraulic properties,including soil water characteristic curve and the unsaturated hydraulic conductivity function,are the key information necessary for study of water and solute movement in the vadose zone.Unfortunately,the great spatial variability and the complexity of soil make direct measurement of these properties costly,time consuming and error-prone.For a large scale application,this may be even infeasible.Hence,indirect methods,which relate more easily measured soil physical properties to hydraulic properties,have been considered valuable alternatives.In the present study,three fractal models, i.e.,Tyler-Wheatcraft,Rieu-Sposito and Brooks-Corey models are used to figure out soil water characteristic curve based on part icle size distribution data.Applicability of these models is evaluated based on the particle-size distribution and soil water characteristic curves of 554 soil samples cited from the Unsaturated Soil Hydraulic Database,UNSODA and other publications.Soil textures of these samples cover the range from sand to heavy clay,and each sample has more than six pairs of pressure-water content measurements.Goodness-of-fit between measured and est imated volumetric water content is expressed in terms of root mean squared error.Results show that,the predictive accuracy of Brooks-Corey model is higher than that of the other two fractal models.In addition,it is suggested that models should be applied to soils texture specifically.In other words,for medium- and coarse-textured soils,Brooks-Corey model is preferable,while Rieu-Sposito model is much more suitable for fine-textured soils.And Tyler-Wheatcraft model is in between the two aforementioned models in terms of estimation error.Tyler-Wheatcraft model can be used for both coarse- and medium-textured soils.