This work aimed to test the ability of laboratory experiments to predict long term changes in plant-available soil P in two grassland experiments carried out on a Luvisol (Ercé, French Pyrenees) and a Brunisol (Gramond, Massif Central, central France) since 1999 and 1998, respectively. Each experiment consisted of two annual rates of N and P fertilization arranged in four randomised blocks. Dry matter yields and P contents were measured for each cut, and P off-take and P balance were calculated every year and accumulated over years of experimentation. The ionic P concentration in solution (Cp) was also determined almost every year for both sites. Results showed that a close linear correlation existed between Cp values and the cumulative P balance. Plant-available P was assessed in laboratory experiments (soil suspension (1 g:10 mL) at steady-state) as the sum of the amount of P ions in soil solution (Pw) plus the associated gross amount (Pr) of P ions bound to the soil solid phase that buffers the soil solution P by diffusion over time (t). The Pw values were calculated by multiplying the P ion concentration in solution (Cp) by the volume/mass ratio. The isotopic labelling of P ions in solution and the dilution principle were used to determine Pr values. The dynamics of Pr versus (Cp, t) was accurately described by the Freundlich kinetic equation (Pr = , with Pr < total mineral P). The Pr values were greater for the Luvisol than for the Brunisol. This behaviour was mainly ascribed to differences in Al- and Fe-(hydr) oxide contents, which were 545 and 117 mmol kg−1 for the Luvisol and Brunisol, respectively. The (Pw + Pr) values were used with the mass conservation law to calculate the change in Cp per unit of P balance (ΔCp) assuming different time scenarios. For the Luvisol, the field and laboratory ΔCp values did not vary significantly, but a mismatch was observed for the Brunisol and this is discussed. Finally, the approach seems promising to elaborate a mechanism- and process-oriented modelling of the changes in plant-available P with management practices in grassland soils.< Réduire