A variety of tests for available soil phosphorus (P) is used worldwide, each with a historical and local background. The test ranging from intensity based ones, indicating P solubility to quantity based ones, i.e. near complete total soil P extraction. A study was set up to identify the most suitable soil P test in a fully comparative way by using the same set of soils of contrasting fields on a set of soil P tests. The requirement for this is to find P response trials in which various fertilizer responses are only related to varying P doses.Soil samples (n=254) were collected from the plough layer of 12 fields complying to these conditions. The soils were collected from the field or from archived soil collection and were gathered in Belgium (2), France (3), Germany (1), United Kingdom (5) and Sweden (1). In all these fields different P fertilisation doses during several years (3-123 years) resulted in differences in soil P status and a significant P response in all 12 fields, sometimes for different crops per field. The relative response, i.e. the crop yield relative to that in highest P rate, ranged between 6.5 and 107.1%. Soil P availability was measured in the 254 gathered soil samples by different methods: (1) extraction with ammonium lactate and acetate at pH 3.75 (Egnér et al., 1960); (2) extraction with 0.5 M NaHCO3 (Olsen et al., 1954); (3) extraction with 0.01 M CaCl2 (Houba et al., 2000); (4) extraction with ammonium oxalate (Schwertmann, 1964), with possibility to calculate the phosphate saturation degree(5) diffusive gradient in thin film technique (DGT) (Degryse et al., 2009). For each method, the relative yield, i.e. the obtained yield divided by the maximum yield observed on the same field, was plotted versus the measured soil P availability. A Mitscherlich curve was fitted on all data and on the data of every field separately. The “critical soil P availability” was defined as that concentration related to 95% relative yield. The best soil P test was determined based on several indices: R² of the Mitscherlich curve, coefficient of variance (CV) of the field specific critical values, CV of the crop type specific critical value, the amount of fault positive or fault negative results included, ... The performances of the tests are approximately equal and differences between the soil P tests are minimal. A more detailed statistical analysis still need to be done, and the results of this analysis will be presented during the conference.< Réduire