Grading has long been recognised to critically influence the mechanical behaviour of ballast. To identify the ballast grading for heavy-haul rail track, monotonic and cyclic triaxial tests are conducted to assess the performances of different gradings. Permanent deformations, aggregates degradation, resilience, shear resistance, maximum and minimum densities are recorded and analysed. The grading is found to affect the behaviour of ballast in that coarser gradings exhibit relatively better strength, resilience and therefore less permanent deformation. However, ballast degradation increases with the overall aggregate size. Therefore, to identify the grading for ballast with different performance objectives, a grey relational theory is used to convert the multi-objective into single-objective, i.e. grey relational grade. A relatively optimal grading that provides the highest grey relational grade is thus suggested for the improved ballast performance.