Accurate and rapid prediction of the spatial structure of soil nitrogen (N) supply would have both economic and environmental benefits with respect to improved inorganic N fertiliser management. Yet traditional biochemical indices of soil N supply have not been widely incorporated into fertiliser decision support systems or environmental risk monitoring programs. Here we illustrate that in a low-input, semi-arid environment, potentially mineralisable N (PMN, as determined by anaerobic incubation) explained 21% of wheat grain yield (P = 0.003), whereas there was no significant relationship between wheat grain yield and inorganic N fertiliser application. We also assessed the spatial pattern of PMN using a structured grid soil sampling strategy over a 10-ha area (180 separate samples, 0–0.1 m). PMN in each soil sample was determined by standard biochemical analysis and also predicted using a fourier transform infrared spectrometer (FTIR). Findings illustrate that FTIR was able to significantly predict (P < 0.001) PMN values in soil and has the advantage of enabling high sample throughput and rapid (within minutes) soil analysis. Given the relatively low cost of FTIR machines and ease of use, such an approach has practical application in situations where analysis cost or access to equipped laboratories has hindered the measurement and monitoring of soil N supply within paddocks and across regions.