ELM mitigation via pacing and plasma fuelling are the two main roles of pellet injection in ITER. ELM triggering by pellet injection provides a promising method for reducing the ELM size as demonstrated on ASDEX Upgrade (AUG). Thus, the ELM impact on plasma-facing components will be minimised. This should be achieved with the smallest possible impact on all other plasma parameters (besides ELM frequency). Hence, pellet injection for ELM pacing should use the smallest possible pellets and a shallow injection. An optimized use of pellets in ITER requires their integration in ITER plasma scenarios in an effective manner for both fuelling and ELM pacing. However, a successful scaling to ITER from results of present day tokamaks requires sound physics understanding. For this purpose, detailed investigations of the local impact of the pellet imposed perturbation are performed at both AUG and JET. For fuelling purposes, pellet injection has to maintain the plasma density and allow high density operation with minimum detrimental effect on energy confinement. This requires pellet penetration as deep as possible into the core plasma. Investigations on JET are under way to identify the best approach to achieve this goal, e.g. by comparing injection from the torus inboard to outboard launch. While the first approach benefits - in hot plasmas - from strong drift effects, the latter has the potential for higher launch speed.