A simple model is proposed which predicts the relative unsaturated hydraulic conductivity function of nonswelling soils by using the first two moments of their water retention curve (WRC). The model is based on the statistical approach but uses less restrictive assumptions concerning the pore configuration than Mualem's <1976> model. The result is that the relative hydraulic conductivity (RHC) is a power function of the relative contribution of the pores filled with water. It is shown that the power value is related to the coefficient of variation characterizing the retention curve, expressed in terms of the WRC model of Assouline et al. <1998>. Therefore a relationship is established between the RHC and the soil structure and texture, as reflected by the measured WRC. The model is calibrated on data from eight soils and tested on data from five soils, representing a wide range of soil textures, from sand to silt. The performances of the model are compared to those resulting from the application of Mualem's <1976> model to the analytical expressions of Brooks and Corey <1964> and Assouline et al. <1998> for the WRC. In most of the cases the proposed model improves the fit of the predicted RHC to the measured data, although its performance in the case of loam soils seems to be weak. ¿ 2001 American Geophysical Union