Soil phosphorus (P) dynamics are of agronomic and environmental importance. Near-infrared (NIR) and mid-infrared (MIR) soil spectroscopy offer rapid and cost-effective alternatives to traditional wet chemical tests. Testing soil P sorption capacity using traditional isothermal sorption experiments allows for the establishment of reference values and the definition of behavioural classes for integration into agronomic guidelines. The aim of this study was to develop infrared spectroscopy as a practical alternative, advancing from laboratory-based to handheld on-site testing of P sorption capacity, with reference to the refined Soil Index System that integrates soil P sorption characteristics to support agronomic decision-making for long-term, site-specific sustainable P management. Mineral topsoil samples (loss-on-ignition ≤ 20%, depth 0.05 - 0.20 m), covering 35,716 km2 of Ireland, were taken from the archive of the Tellus national programme. Calibration samples were characterised by P status and Langmuir sorption parameters, including P sorption maximum capacity (Smax, mg·kg⁻¹) and binding affinity from the wet chemical extractions. Segmented regression identified a significant change-point in P behaviour at Smax = 450.03 mg·kg−1, classifying soils into ‘low’ (SL) and ‘high’ (SH) P sorption capacities. Chemometric models for Smax prediction were developed using samples in a dry state. Benchtop (Bruker, 4000–400 cm⁻¹) and handheld (Agilent, 4000–650 cm⁻¹) MIR spectrometers were used to scan <0.100 mm (ball-milled) and <2 mm soil samples. In comparison, Agilent obtained spectral libraries contained more noise and less resolution. Unlike Bruker, ball milling preparation step significantly influenced the performance of Agilent. Handheld NIR instruments (InnoSpectra, 900–1700 nm, and NeoSpectra, 1350–2550 nm) were also evaluated for <2 mm samples. Only Agilent and NeoSpectra handheld instruments demonstrated potential for discriminating SL/SH soils. A systematic re-wetting experiment assessed moisture effects on <2 mm samples scanned with handheld NIR and MIR spectrometers. The External Parameter Orthogonalisation (EPO) algorithm successfully removed moisture effects. However, predictions from EPO-transformed wet spectra remained less accurate than those from dry samples. In practice, to simplify and improve the reliability of predictions from MIR spectroscopy, maintaining relatively dry field conditions is highly recommended. The findings in this thesis contribute to site-specific P management, water quality protection, and policy development for future studies.