The ability to store and remobilise large amounts of stem water soluble carbohydrates (WSC) is a desirable trait to incorporate when breeding wheat for environments where terminal water stress occurs frequently. The trait is under complex genetic control (many independent quantitative trait loci of individual small effect), limiting the direct use of marker assisted selection in breeding programs. This highlights the need for a high throughput phenotyping method. The aim of this study was to evaluate if it was possible to estimate the concentration and amount of stem WSC in wheat canopies non-destructively using a hyperspectral sensor, as opposed to currently available labour intensive methods involving harvesting, processing, grinding and laboratory protocols. The method was developed using recombinant inbred lines, contrasting for WSC but similar for height and phenology, under a range of management conditions. WSC concentration and amount were successfully (R²>0.85) predicted using partial least square regression in the spectral range from 350 to 1300 nm.