Miscible displacement experiments were conducted on five soils using 7.5-cm diam and either 2.4- or 5.1-cm long columns. Glycerophosphate (771-2,195 ppm-P) and chloride (100 meq/liter) were applied in a 50-ml pulse to soil columns maintained at steady water contents slightly less than saturation. Two mathematical models were used in fitting the effluent concentration curves. The first was an analytical solution which consisted of first-order hydrolysis kinetics and reversible, instantaneous, linear adsorption reactions. The second was an explicit-implicit finite difference solution which consisted of first-order hydrolysis kinetics and two types of adsorption sites, one of which was reversible, instantaneous, and nonlinear; the other was reversible, kinetic, and nonlinear. The values of the partition coefficient and the exponent of the Freundlich isotherm used in the two-site model were determined by batch adsorption isotherm experiments. The rate coefficients of the adsorption-desorption reactions and hydrolysis reactions as well as the proportions of adsorption sites that were kinetic or equilibrium were determined by fitting calculated to measured glycerophosphate elution curves. Both the linear adsorption partition coefficient and the hydrolysis rate coefficient were determined by fitting elution curves of glycerophosphate using the linear, equilibrium model. The fitted partition coefficient was not in agreement with the batch adsorption isotherm indicating that the adsorption-desorption reactions were not instantaneous. Both mathematical models were capable of fitting the chloride effluent curves of all five soils quite well. The two-site model was capable of fitting the glycerophosphate effluent curves of all five soils more closely than the linear, equilibrium model. The two-site model, though, was incapable of accurately describing the adsorption-desorption reactions in two of the soils. This was apparently due to a wide range of kinetic rates of the reactions in these two soils.