The efficiency of a partially acidulated rock phosphate, FAPS, was studied using bean as a test crop in a Dark Red Latosol low in P and acidic. For comparison ground rock phosphate, from Araxá, MG, Brazil (FA) and from Gafsa (FG) , and simple superphosphate, either powder or granulated (SPSP and SPSG, respectively) were employed. Three successive croppings were conducted in order to estimate the residual effect of the P2O5sources. As subtreatments initial liming and no liming were used. The statistical analyses of the data allowed for the following conclusions to be drawn, taking into account the dry matter production observed in the three harvests, and making the yield obtained at the expenses of SPSP equal to 100: (1) in the absence of liming - SPSG = 100, FA = 60 to 85 (depending upon granule size), FG = 132; (2) in the presence of liming - SPSG = 105, FA = 52, FAPS = 75 to 90, FG = 95. Irrespective of source of P2O5 liming increased the absorption of P only on the third crop. FAPS acted also as source of Ca and S. Liming decreased residual available (Olsen) P determined after the third cropping except in the case of the treatment with larger granules. The available residual P obeyed the following decreasing order: SPS, FAPS, FG and FA. It is considered that FAPS behaved in a way closer to that of the SPS.