Known amounts of phosphate were added to a series of soils and, after incubation at constant temperature, desorption was measured by shaking the soils with a range of volumes of phosphate-free solutions. When the method of shaking was vigorous, an initial desorption of phosphate was followed by an apparent resorption. Since this did not happen when shaking was gentle, it was thought to be due to breakdown of soil particles and exposure of new surfaces. When gentle shaking was used, desorption was almost complete within a few hours when the solution:soil ratio was small but continued up to at least 96 hr when the ratio was large. The amount of phosphate which could be desorbed decreased as the period of prior contact between soil and phosphate increased. The rate and amount of desorption could be described by an equation which indicated that the proportion of the phosphate which could be desorbed increased with a fractional power of the period of desorption, and decreased with a fractional power of the period of prior contact. The amount of desorption also depended on intrinsic soil properties and it was found that, for a range of soils, a parameter reflecting desorption was highly correlated with a measure of adsorption.