In Brazil, where the rock phosphates are high in impurities, no attempt has been made to evaluate the P supplying efficiency of the neutral ammonium citrate fraction (NAC) of P fertilizers, or to verify if the NAC + H2O extraction solution (AOAC) is satisfactory for estimating the P availability. To attain these objectives, a greenhouse experiment was carried out with samples of a Typic Hapludox soil. Four acidulated phosphates obtained from Brazilian raw materials were studied; monocalcium phosphate p.a. [Ca(H2PO4)(2). H2O] was included as a standard source of P, as well as leached samples containing no water-soluble P. The fertilizers were thoroughly mixed with the whole soil in the pots or with only 1% of its volume, at the rate of 50 mg kg(-1) of P, soluble in NAC + H2O. Corn plants (Zea mays, L.) were grown for 35 days and the amounts of dry matter and P accumulated in plant tops were determined. Increasing the amount of cationic impurities in the raw materials decreased the concentration of water-soluble P, NAC + H2O-soluble P and water-soluble P/NAC + H2O soluble P ratio of the fertilizers obtained. The P in the NAC fraction was not as much available to plants as in the NAC + H2O fraction or in pure MCP. The great variation found in dry matter (5.4 to 17.1 g pot(-1)) and in P uptake (6.3 to 22.2 mg pot(-1)) indicates that the AOAC method is not an adequate index for evaluating the P availability of fertilizers with high amounts of cationic impurities.