In the process of phosphate rock acidulation, several impure P compounds may be formed along with the desirable Ca and NH4 phosphates. Such compounds normally reduce the content of water-soluble P and thus the agronomic effectiveness of commercial fertilizers. In order to study this problem, a greenhouse experiment consisting of three consecutive corn crops was conducted in samples of a Red-Yellow Latosol (Typical Hapludox) in a completely randomized design (6 x 2 x 2), with four replicates. Six commercial fertilizers were added to 2 kg of soil at a rate of 70 mg kg-1 P, based on the content of soluble P in neutral ammonium citrate plus water (NAC + H2O) of the fertilizers. Fertilizer application occurred either in the original form or leached to remove the water-soluble fraction, either by mixing the fertilizer with the whole soil in the pots or with only 1 % of its volume. The corn plants were harvested 40 days after emergence to determine the shoot dry matter and accumulated P. For the first crop and localized application, the elimination of water-soluble P from the original fertilizers resulted in less bioavailable P for the plants. For the second and third crops, the effects of P source, leaching and application methods were not as evident as for the first, suggesting that the tested P sources may have similar efficiencies when considering successive cropping. The conclusion was drawn that the water-insoluble but NAC-soluble fractions of commercial P fertilizers are not necessarily inert because they can provide P in the long run.