The growing concerns about water eutrophication have made it urgent to restrict losses of phosphorus (P) from agricultural soils and to develop methods for predicting such losses. In this work, we used the paradigm of P sorption-desorption curves to confirm the hypothesis that the amount of dissolved reactive phosphorus (DRP) released to a dilute electrolyte tends to be proportional to the concentration of DRP in the soil solution raised to a power that decreases with increasing solution to soil ratio (W). The hypothesis was tested for a group of 12 widely ranging European agricultural soils fertilized with P in excess of crop needs. Phosphorus desorption was studied under near-static and turbulent conditions in laboratory experiments. The concentration of DRP in the 1:1 soil to water extract (P1:1) was used as a proxy for the DRP concentration in the soil solution. The amount of desorbed P was found to be correlated with P1:1 raised to a power that decreased from 0.7 to 0.9 at W=100 to 0.2 to 0.4 at W=10 000. Correlation was not improved by introducing additional variables related to P sorption-desorption properties. Olsen P was found to be of lower predictive value than P1:1. Also, the index of degree of soil saturation with phosphorus (DSSP) based on oxalate extraction failed to predict P desorption. The fact that P1:1 seemingly predicts P desorption accurately for a wide range of soils makes it potentially useful in areas of high soil diversity.