Stochastic modeling of soil water fluxes in the absence of measured hydraulic parameters requires a knowledge of the expected distribution of the hydraulic parameters in different soil types. Predictive relationships describing the hydraulic parameter distributions must be developed based on the common descriptors of the physical properties of soils (e.g., texture, structure, particle size distribution). Covariation among the hydraulic parameters within these relationships must be identified. Data for 1448 soil samples were examined in an evaluation of the usefulness of qualitative descriptors as predictors of soil hydraulic behavior. Analysis of variance and multiple linear regression techniques were used to derive quantitative expressions for the moments of the hydraulic parameters as functions of the particle size distributions (percent sand, silt, and clay content) of soils. Discriminant analysis suggests that the covariation of the hydraulic parameters can be used to construct a classification scheme based on the hydraulic behavior of soils that is analogous to the textural classification scheme based on the sand, silt, and clay content of soils.