Phosphate (P(i)) release due to Fe(III) oxide dissolution is well documented for soils undergoing reduction. The P(i) sorption properties of soils in anoxic conditions are, however, still under consideration. In this investigation, P(i) sorption to strictly anoxic soils was compared with oxic conditions to assess the potential of lowland soils to function as traps for P(i) when flooded with drainage water. Batch sorption experiments were performed on seven minerogenic soils. Sorption to the anoxic soils was conducted after anoxic incubation, resulting in reduction of 36 to 93% of the dithionite-extractable Fe(III) (Fe(BD)). Langmuir fitted P(i) sorption isotherms showed a P(i) release of up to 1.1 mmol kg(-1) in six soils when P(i) concentrations in the matrix (P(sol)) were lower than 10 microM. Phosphate desorption was attributed to dissolution of amorphous iron oxides, and higher pH under anoxic conditions. The point of zero net sorption (EPC(0)) increased 2- to 10-fold on reduction. Five soils showed higher P(i) sorption capacities in the anoxic than in the oxic state at higher P(sol) concentrations. Solubility calculations indicated that precipitation of vivianite or similar Fe(II) phosphates may have caused the higher sorption capacities. Use of maximum sorption capacity (S(max)) is therefore misleading as a measure of P(i) sorption at low P(sol) concentrations. The results demonstrate that none of the strongly anoxic soils, irrespective of the initial Fe(III) oxide content, the P saturation, and the degree of Fe(III) oxide reduction, could retain P(i) at natural P(sol) concentrations in agricultural drainage water.