Samples were collected from 228 soils from the main agricultural areas of Western Australia (W.A.) and their chemical and textural properties analysed. Soils were incubated for 7 days with North Carolina rock phosphate at a rate equivalent to 1022 mu g Ca g(-1) and 383 mu g P g(-1) soil. Subsequently, the amount of rock phosphate (RP) that had dissolved was determined by measuring the increase in extractable calcium (Delta Ca) and changes in plant-available P were estimated by a sodium bicarbonate extraction (Delta Bic-P). Only 29 soils dissolved more than 40% of the added RP and these occur in the wettest area of W.A., to the west of the 800 mm rainfall isohyet. They represent a variety of soil types including Vertisols, Spodosols, Alfisols and Ultisols. Simple and multiple regressions were carried out to investigate the soil properties responsible for RP dissolution and associated P availability. No single soil property adequately predicted RP dissolution or concomitant changes in bicarbonate-P. Titratable acidity and sand content together explained 61% of the variance in RP dissolution; titratable acidity was the most predictive single variable for bicarbonate-P but only explained 56% of the variance. Stepwise regression showed that titratable acidity and pH((CaCl2)) together explained 79% of the variance in bicarbonate-P. The most suitable soils for RP application are sandy, humic or peaty podzols where much of the dissolved P remains available to plants. The other soils (mainly red and yellow earths and Vertisols) which dissolved much RP contain larger amounts of extractable aluminium (by dithionite or oxalate) which is the soil component largely responsible for P-sorption in these W.A. soils. These results indicate that RP fertilizers will be most effective on the humic, sandy podzols of the western and southern coastal plains of WA.