ABSTRACT Assessment of phosphorus (P)-infiltrating croplands is essential for the preservation of the water environment. It has been pointed out that a huge discrepancy lies in the different evaluation methods of P adsorption, such as batch experiments and column experiments, which makes it difficult to demonstrate P mobility under flow conditions. The objective of this study was to evaluate the applicability of the convective-dispersion equation using the parameters of the Langmuir-type isotherm obtained from batch experiments with different reaction times: the adsorption capacity of soil (q max) = 0.112 (g kg−1) for a Gray lowland soil with 24 h reaction time, q max = 0.484 (g kg−1) for an Andosol (volcanic ash soil) with 24 h reaction time, and q max = 1.17 (g kg−1) for an Andosol with 32 d reaction time, for describing P mobility in typical Japanese agricultural soils under fast flow conditions. The breakthrough curves of P infiltrating the soil columns demonstrate nonequilibrium P adsorption by the soil. The chemical nonequilibrium model, with a kinetic adsorption rate of α = 0.40 (Gray lowland soil) and 0.098 (Andosol), succeeded in describing the observations in the column experiments. Compared with Gray lowland soil, which is relatively rich in iron oxide, P mobility was largely affected by kinetic sorption in Andosol, which is relatively rich in allophane. It is suggested that the P adsorption capacity of soils should be evaluated reflecting the soil composition in order to simulate the P mobility under flow conditions. In particular, the slow adsorption (long-lasting adsorption) of P by the soil should be considered in the estimation of the P transport.