The Mehlich-1 method is the most used in Brazil to extract phosphorus (P) available in the soil. Based on the hypothesis of loss of the predictive capacity of this extractant in alkaline soils, the Mehlich-1, Olsen and ion exchange resin methods were compared in samples of ten representative soils from the Brazil’s semi-arid region, between the Piranhas-Açu (RN) and Jaguaribe (CE) valleys, with soil pH ranging from 4.1 to 8.0. The soil samples received four doses of P according to their phosphorus buffer capacity (PBC) and were incubated for 30 days. The treatments were distributed in randomized blocks, in a 10 × 4 factorial arrangement, with ten soils and four doses of phosphorus, with three replications. Each experimental unit consisted of a plastic pot containing 3.0 dm³ of soil, where two corn plants were grown for 31 days in a greenhouse. The P extracted by the methods was correlated with the P accumulated in the aerial part of the corn plants. The mixed resin and Olsen methods had the best correlations with the accumulated P in the plants, regardless of the pH and the PBC of the soils. The acidity depletion in Mehlich-1 extracts was significant in alkaline soils, with a high CaCO3 equivalent content. Mehlich-1 was suitable in acidic and low PBC soils, but in soils with higher pH and Ca2+ content, this extractant may overestimate the P available in soils naturally rich in P-Ca or underestimate the P available in soils rich in CaCO3