Coarse granular materials such as rockfill have been described in recent years by means of strain and work hardening plasticity models Oldecop and Alonso (2001), Chavez and Alonso (2003), Alonso et al. (2005) which require a proper definition of the flow rule. The paper presents experimental information oil dilatancy rates obtained in suction controlled triaxial tests performed on large diameter (25 cm) specimens of compacted coarse gravels of hard limestone. Tests involved confining stresses in the range 0.3-1 MPa and Relative Humidity in the range 10-100%. Unloading-reloading during the tests provided information on the elastic response. Volumetric and deviatoric plastic strains rates were determined from records of total vertical and volumetric deformations. In general, a steady (critical) state was not found. Plots have been prepared showing the variation of the plastic flow direction, d, with a number of variables: vertical deformation, stress ratio, eta, confining stress, p, plastic work, W-p, suction, s, and some derived normalized variables. It was found that the common framework of dilatancy being a function of the current stress ratio and the limiting stress ratio is far from being appropriate to describe the behaviour of the tested aggregates. It was found that relationships d = f (eta W-p/p, s), provide a good description of test results. The paper discusses the rationale behind this selection.