Effects of soil organic matter (SOM) on P sorption of soils still remain to be clarified because contradictory results have been reported in the literature. In the present study, pH-dependent P sorption on an allophanic Andisol and an alluvial soil was compared with that on hydrogen peroxide (H(2)O(2))-treated, acid-oxalate (OX)-treated, and dithionite-citrate-bicarbonate (DCB)-treated soils. Removal of SOM increased or decreased P sorption depending on the equilibrium pH values and soil types. In the H(2)O(2)-, OX-, and DCB-treated soils, P sorption was pH-dependent, but this trend was not conspicuous in the untreated soils. It is likely that SOM affects P sorption of soils through three factors, competitive sorption, inhibition of polymerization and crystallization of metals such as Al and Fe, and flexible structure of metal-SOM complexes. As a result, the number of available sites for P sorption would remain relatively constant in the wide range of equilibrium pH values in the presence of SOM. The P sorption characteristics were analyzed at constant equilibrium pH values (4.0 to 7.0) using the Langmuir equation as a local isotherm. The maximum number of available sites for P sorption (Q(max)) was pH-dependent in the H(2)O(2)-, OX-, and DCB-treated soils, while this trend was not conspicuous in the untreated soils. Affinity constants related to binding strength (K) were less affected by the equilibrium pH values, soil types, and soil treatments, and were almost constant (log K approximate to 4.5). These findings support the hypothesis that SOM plays a role in keeping the number of available sites for P sorption relatively constant but does not affect the P sorption affinity. By estimating the Q(max) and K values as a function of equilibrium pH values, pH-dependent P sorption was well simulated with four or two adjustable parameters. This empirical model could be useful and convenient for a rough estimation of the pH-dependent P sorption of soils.