Soil samples collected from 8 field experiments in Western Australia to which 5 - 8 amounts of superphosphate had been applied once only 13 - 23 years previously were used to measure the phosphorus ( P) buffer capacity of soil (PBC) and P sorption by several single-point indices. PBC was estimated from well-defined P sorption curves when several levels of P were added to soil suspensions, and was the amount of P sorbed when the concentration of P in the final solution was raised from 0.25 to 0.35 mg P/L. The single-point P sorption indices were measured by adding one amount of P ( 10 mg P/L) to soil suspensions (1 : 20, soil : 0.02 M KCl or 0.01 M CaCl2). Three indices were calculated from the amount of P sorbed by soil (S, mg P/kg soil) and the amount of P in solution (c, mg P/ L) -(1) the phosphorus retention index ( PRI, S/c [L/kg]), (2) the Freundlich retention index (FRI, S/c(0.35) [dimensionless]), and (3) the phosphorus sorption index (PSI, S/ log(10) [c x 1000] [ dimensionless]) - to provide PRI (K & Ca), FRI (K & Ca), and PSI (K & Ca) values. P sorption was also measured by the P buffer index (PBI), the new single-point P sorption index recommended for national use, to provide PBICa values. To estimate the previous P sorbed by soil ( native soil P is negligible for these soils, so previously sorbed P originates from fertiliser P applied in a previous year), the amount of P extracted by 0.5 M sodium bicarbonate from soil (Colwell soil test P) was added to the amount of P sorbed by soil to calculate PRI* (K & Ca), FRI* (K & Ca), PSI* (K & Ca), and PBI* (Ca) values. In addition, previously sorbed P was estimated using the q coefficient of the Freundlich equation; q was added to P sorption to calculate PSI**, FRI**, PSI** and PBI** values to take account of previously sorbed P. For the 8 experiments, PBC values significantly decreased where more fertiliser P had been applied to the soils 13 - 23 years previously. This indicated that the capacity of the 8 soils to sorb P decreased as more P was applied in a previous year, and a single-point P sorption index would need to reflect this decrease. As the amount of P applied to soil in the field plots increased, the following trends occurred: ( 1) Colwell soil test P always increased; (2) PRIK & Ca, FRIK & Ca, PSIK & Ca, and PBICa consistently decreased; (3) PRI* (K & Ca), FRI* K & Ca, PSI* K & Ca, and PBI* Ca mostly increased, but with some values being unaffected or decreasing; (4) PRI**, FRI**, PSI**, and PBI** values were largely unaffected by the amount of P applied in a previous year. Evidently, either adding Colwell soil test P or q to P sorption to calculate the single-point P sorption indices mostly overestimated P sorption by the sandy, low P sorbing soils used, but the overestimate was larger for Colwell soil test P than for q.