Soil samples were collected from 14 long-term field experiments in south-western Australia to which several amounts of superphosphate or phosphate rock had been applied in a previous year. The samples were analysed for phosphorus (P) by the Colwell sodium bicarbonate procedure, presently used in Western Australia, and the Mehlich 3 procedure, being assessed as a new multi-element test for the region. For the Mehlich procedure, the concentration of total and inorganic P in the extract solution was measured. The soil test values were related to yields of crops and pasture measured later on in the year in which the soil samples were collected.The Mehlich 3 procedures (Mehlich 3 total and Mehlich 3 inorganic soil test P values) were similar, with the total values mostly being slightly larger. For soil treated with superphosphate, for each year of each experiment: (i) Mehlich 3 values were closely correlated with Colwell values; and (ii) the relationship between plant yield and soil test P (the soil P test calibration) was similar for the Colwell and Mehlich 3 procedures. However, for soil treated with phosphate rock, the Colwell procedure consistently produced lower soil test P values than the Mehlich 3 procedure, and the calibration relating plant yield to soil test P was different for the Colwell and Mehlich 3 procedures, indicating, for soils treated with phosphate rock, separate calibrations are required for the 2 procedures. We conclude that for soils of south-western Australia treated with superphosphate (most of the soils), the Mehlich 3 procedure can be used instead of the Colwell procedure to measure soil test P, providing support for the Mehlich 3 procedure to be developed as the multi-element soil test for the region.