The analysis of two well-diagnosed JET H-mode pedestals using MHD and gyro-kinetic codes shows that the pedestals are close to the stability limits of the peeling-ballooning modes at the end of the ELM cycle, but no unstable kinetic ballooning modes (KBM) are found in the steep pedestal region during the ELM cycle. Good correspondence between ideal MHD n = ballooning mode and gyro-kinetic KBM stability is confirmed. The top of the pedestal is found to have unstable micro-tearing modes that are sub-dominant to ion temperature gradient modes. Stability analysis of the pedestal used in ITER transport simulations shows that a pedestal temperature of 5.9keV (corresponding to 107kPa with ne = 0.85nGW) is at the stability limit of a Type I ELMy H-mode pedestal limited by peeling-ballooning modes. The close fitting metal wall in ITER is shown to have the potential to affect the ideal peeling-ballooning instabilities responsible for triggering ELMs, thus modifying the dynamics of the ELM crash.