We have carried out nonequilibrium molecular dynamics simulations of the compaction of a single three-dimensional granule composed of over 1000 Lennard-Jones (LJ) particles. The granule was contained within an orthorhombic box with repulsive walls and deformed by a vertically moving top wall. The compaction cycle adopted was intended to mimic the procedure employed in industrial tabletting processes, by compressing the granule during the downward movement of the top wall (compaction) followed by an upward movement of the top wall (decompaction). We have explored the effects of different compression rates on the deformation, microstructure, and the final integrity of the granule. Although the simulations are formally atomistic, we believe a mesoscopic significance can be attached to the results that makes them relevant to the larger scale compaction involved in industrially relevant processes. The cluster representation of the granule allows for significant deformation during the process, and the simulations reproduce a number of well-known effects found in the pharmaceutical tabletting and other literature. Rapid compaction resulted in an essentially elastic response and even break up of the formed tablet during the decompaction stage, an effect known as lamination. Slower compaction speeds, which enabled greater internal rearrangement of the LJ particles through plastic deformation, produced a more structurally uniform tablet at the end of the cycle. For the faster compaction speed the top wall moved away faster than the compacted material could recover, giving rise to misleadingly low values of the apparent elastic response of the material as measured by the force from the material on the top wall. We believe this could be an important issue when interpreting experimental data. These simulations were able to capture the transition between the fast and slow compaction rate regimes and reveal some rudiments of the lamination problem that plagues the industrial process of tabletting. (C) 2003 American Institute of Physics.