After 32PO4 labelling of the isotopically exchangeable phosphate ions of soils, « available soil phosphorus » was extracted by different chemical reagents used for soil routine analysis, i.e. sodium bicarbonate for the OLSEN method, ammonium oxalate for the Joret-Hébert method, citric acid for the DYER method and CaCl2 solution. Phosphate ions and 32PO4 ions were determined in the various extracts. They were compared to those obtained in the soil-solution where isotopic dilution kinetics were performed. The specific radioactivities (RAS) were calculated and compared. Because crops, even with mycorrhizae, take up their phosphorus from soils only as phosphate ions in the soil solution, we have considered that the better methods were not those where a maximum of 32PO4 was extracted but the methods where the RAS of the extracted P was close to the RAS of the soil-solution phosphate. With this condition the best method was the OLSEN method, not only for calcareous soils but also for acid soils. From these experiments it can be also deduced that, due to the interactions between soils and extractants, it is not possible to find a universal method for extraction in all soil types. For a better understanding of available soil phosphorus and for better agricultural practice of P fertilization it seems necessary to take into account the three following parameters : quantity, capacity and intensity.