Previous research on random fibre reinforced granular materials has shown that the relative dimensions of the grains and fibres significantly affect the macro-mechanical behaviour of the mixture. However, quantitative data are scarce and most previous work has focused on fine to medium sands, leaving uncertainties regarding the applicability of current knowledge to larger size aggregates such as railway ballast. In this paper, triaxial test data on 1 /3 and 1 /5 scale railway ballast are used to develop scaling relationships for the size and quantity of fibres needed to achieve the same reinforcing effect in granular materials of differing grain size. It is shown that, to maintain consistency across scales, fibre content should be quantified as a numerical (i.e. number of fibres per grain) rather than a volumetric ratio. It is further shown that increasing the fibre length increases the resistance of the mixture to deviator stress if the fibres are wide enough; and that provided an allowance is made for the effect of fibre tension, the changes in the stress-strain-strength behaviour of the granular matrix resulting from the changes in void ratio associated with the addition of the fibres are consistent with conventional soil mechanics theory across scales.