The common soil-P tests are unable to distinguish between P-deficient and P-sufficient Sri Lankan soils used for rice cultivation, therefore adsorption experiments were performed to determine whether sorption parameters as obtained from the Tempkin equation are more appropriate. Phosphate sorption is mainly governed by amorphous Fe- and Al-hydroxides and (or) oxides, clay content and its mineralogy, soil carbon content, and the amount of gibbsite. Removal of part of the available P with the Pi-P test prior to the sorption experiment, affected the so-called P sorption maximum, a but not the soil buffering capacity, b. Soil extraction with oxalate reduced both sorption parameters (a and b) significantly. A negative correlation was obtained between Tempkin sorption coefficients and the per cent shoot P and yield (only for buffering capacity). Using the P-sorption data it can be explained why some fields responded to P and some did not.