Particle size distribution parameters are known to greatly affect the mixing behavior of concrete but no clear mix design rules have been defined to avoid possible mixing difficulties. This article will describe the impact of the size of the individual particles, of the packing properties in simple cases of granular assemblies during the first moments of the mixing operation. The materials used for these experiments were model systems made up of various sizes of glass beads. The experimental equipment was a granulation rheometer used as an instrumented mixer. Different phases were observed for the powder/paste transition and can be compared to those observed in wet granulation conditions. The system went through the various and usual states described in the wet granulation process (pendular, funicular. etc). A signal increase was of during the mixing process and could be related to an increase of the mix's cohesion force; this was greater when the particles were fine. The dilatancy measurements made at the same time showed important inter-particle forces and were high enough to induce a collective behavior of the particles, hence inducing dilatancy levels that reached high values very quickly. These results highlight the important effect of the size of the individual particles on the obtained responses, with a very strong effect on the turning point. Maximum intensity reached on individually characterized beads evolved by 1/d(50). This could very quickly become critical in an industrial context (stalling of the mixer). This result was partially explained by Rumpfs model [H. Rumpf, in W.A. Knepper (ed.), Agglomeration, Interscience, New York (1962) 379]. The maximum value in a binary mix of beads also depends on the particle size characteristics of the mixes. Optimization of the maximum value is possible by using the packing properties. The minimum value was obtained in our systems for approximately 20% of the fine particles. The time needed to reach the paste state was also affected by the composition of the mix. It appeared to depend on the system's degree of saturation, therefore on the packing density of the mix. The usual complementary granular levers could modify the system's void fraction and optimize the time devoted to this initial mixing phase, This is already an interesting result by itself since it is compatible with the minimum void fraction often looked for in mix designs. (C) 2008 Elsevier B.V. All rights reserved.