As the heating is increased through a "power threshold" a tokamak plasma undergoes a bifurcation to a state of high confinement, called the H-mode. This high confinement is due to a thin, insulating region that forms close to the plasma edge, called the pedestal region, which supports steep plasma pressure gradients. The quality of the confinement is determined by how steep the pressure gradient is and the width of the pedestal region - predictions for the fusion performance of ITER are therefore rather sensitive to these pedestal parameters.

In this talk, we will explore the plasma physics that controls the properties of the pedestal region. Micro-instabilities drive turbulence there, which influence the pressure gradient that can be supported. We shall consider the properties of such micro-instabilities and describe some of the challenging features that studying these in a toroidal plasma geometry raises. The steep pressure gradient can also drive larger scale, faster growing (ideal MHD) instabilities. These drive plasma eruptions called Edge Localised Modes, or ELMs, which explode from the plasma surface as filamentary structures, like mini "solar flares". Together with the turbulent transport associated with the micro-instabilities, the ELMs provide a second constraint that yields both the pedestal width and pressure gradient.

ELMs expel large amounts of energy and particles and on future tokamaks, such as ITER, they could cause significant damage (eg through erosion of plasma-facing structures). We will review the physics of ELMs, and describe some of the techniques that are being developed to control them on ITER. Also, on existing tokamaks, a number of "small ELM" regimes have been identified which could provide a solution for ITER. However, their physics mechanism remains uncertain, so understanding whether or not ITER can access these small ELM regimes remains unclear. We will pull together what we have learnt about micro-instabilities and ELMs to suggest a new physics model for small ELMs, that could provide the basis for a predictive capability for ITER.