The sustainable management of phosphorus (P) fertility in agroecosystems depends on either soil- or plant-based diagnosis methods. Our overall objective was to determine the relationship between the P nutrition index based on a critical P dilution of maize (Zea mays. L.) and indicators of the soil P supply assessed using either a process-based approach or the common chemical extraction of the Olsen’ method. Long-term field experiments on mineral P fertilization, conducted at four sites with contrasting soil types representative of the main types of cropped soils in France, were selected to have deficient to excessive plant-available soil P. Three annual rates of superphosphate including no-P (P0) and about one (P1) and two (P2) times the annual P exported in harvests were studied, while the N and K supply was non-limiting. The shoot biomass (W, Mg DM ha−1) and its P concentration (Pmaize, g P kg−1 DM) were determined at 5–8 dates during the plant cycle after more or less two decades of P applications. Plant-available soil P in the ploughed layer was assessed by a functional and process-based approach that consists in determining together orthophosphate ions (oPions) concentration (CP, mg P L−1), and the oPions amount (Qw) in solution, and the amounts of diffusive oPions bound to soils (Pr, mg P kg−1 soil) that equilibrates the solution over time (t in minutes). We also assessed plant-available soil P by the Olsen’ extraction (Olsen-P). Shoot biomass significantly decreased for the P0 treatment while there was no significant differences between the P1 and P2 treatments. The Pmaize decreased as W increased. The criticalPmaize curve, the minimum Pmaize required to achieve maximum growth, was estimated using all data of the P1 treatment from the four sites: 3.66 × W−0.238 (r² = 0.61 for 78 observations). At maturity, the P nutritional index (PNI), calculated as the measured Pmaize divided by the criticalPmaize, ranged from 53 % to 92 % for the P0 treatment and from 94 % to 128 % for the P1 and P2 treatments. The CP values ranged from 0.04 to 2.25 mg P L−1. The Pr values were described accurately by: Pr = v × CPw × tp for t <400 min with (v, w, p) highly different across soils types. Calibrations of PNI to CP or to Olsen-P were both soil-specific. The new result brought by this study was that a unique calibration curve for all sites was obtained for PNI vs (Qw + Pr) considering a period of resupplying soil solution oPions of about one day. All points fell on the same regression line suggesting that this soil- and plant-based methods of diagnossis would both equally effective. This result provides new evidence of much better relevance and reliability of the process-based assessment rather than chemical extraction to adequately estimate the fraction of soil P that contributes to plant nutrition. The plant available soil P stock was better evaluated by the process-based approach because oPions buffering capacity of soils was accounted for. Consequently, the relationship between plant and soil indicators was highly improved.