The initial and residual effectiveness of superphosphate, Christmas Island C-grade ore (C-ore) and 500�C heated (calcined) C-ore (Calciphos) was measured on a deep, very sandy soil near Badgingarra, W.A. Different levels of each fertiliser were applied once only in May 1980. Yields of lupins (1980 and 1983) and subterranean clover (1981) were used to determine the effectiveness of the fertilisers, and in 1986 the residual value of the fertilisers was measured for wheat, oats, and barley, relative to freshly applied superphosphate. Samples of the top 10 cm of soil were collected each January-February for soil tests for P which were compared to plant yields measured in the following spring (Aug.-Dec.). Superphosphate was the most effective fertiliser, though its effectiveness declined by about 50% between years 1 and 2, and by a further 25% between years 2 and 4. C-ore was only about one-tenth as effective as superphosphate in the year of application, and its effectiveness declined by about 40% between years 1 and 2, and by a further 15% between years 2 and 4. Calciphos was about one-fifth as effective as superphosphate in the year of application, and its effectiveness declined by about 40% between years 1 and 2, and remained approximately constant between years 2 and 4. In 1986, the effectiveness of 1980 applied superphosphate was between about one-quarter to one-fifth as effective as freshly applied superphosphate for wheat, oats and barley, and the 1980 applied C-ore and Calciphos were about 15% as effective as freshly applied superphosphate. The amount of P extracted by sodium bicarbonate from soil fertilised with superphosphate decreased by about 50% between years 2 and 3, by a further 20% between years 3 and 4, and by a further 15% between years 4 and 7. For soil treated with Calciphos, the amount of extractable P doubled between years 2 and 3 and thereafter remained approximately constant. Very little P was extracted from soil fertilised with C-ore. Separate calibration curves were required in most cases for the relationship between yield and soil test values for superphosphate and rock phosphate in 1981 and 1983, and for each fertiliser and species in 1986. A substantial amount of superphosphate-added P (50-70%) and about 20% of rock phosphate-added P had leached below 50 cm of the topsoil 6 years after application.