Equations based on the chemical composition of feedstuffs have traditionally been used to predict the energy value of cereal-based diets for pigs. However, the dietary fibre fraction is the greatest source of variation in predicting energy value and it is recognised that the predictive power of equations could be improved by the use of more detailed chemical analysis of the fibre fraction, as opposed to using more traditional gravimetric techniques which do not define specific chemical entities. The potentially fermentable fraction of dietary fibre is the non-starch polysaccharide (NSP) portion and there are methods available to allow the routine measurement of total NSP and its constituent monomers. In this study we investigated the effect of including total NSP and its major constituent monomers as independent variables in data sets for step-wise regression analysis in addition to more traditional measurements of fibre content on the accuracy of equations designed to predict the energy value of feedstuffs for pigs.