Numerical modelling using computational fluid mechanics (CFD) and discrete element method (DEM) becomes increasingly prevalent for the exploration of agglomeration and deagglomeration in dry powder inhalers (DPIs). These techniques provide detailed information on air flow and particle-particle/wall interaction, respectively. Coupling of CFD and DEM enables an in-depth investigation of the mechanisms at the microscopic level. This paper reviews the applications of CFD and DEM in DPI development and optimisation. The recent progress in modelling of two key processes in DPIs, i.e. agglomeration and deagglomeration, is presented. It has been demonstrated that DEM-CFD is a promising numerical approach to investigate the underlying agglomeration and deagglomeration mechanisms for DPIs. With further advances in computing capacity, it is expected that DEM-CFD will be capable of addressing more realistic and complicated issues in DPI improvement.