Neoclassical tearing modes are one of the most serious concerns for operation on a next-step tokamak device. The modes occur on present tokamaks at normalized pressure (beta (N)) values comparable to those envisaged For baseline scenarios in future devices, such as ITER-FEAT. Further, empirical scalings based on data from many of the present machines point to much lower thresholds on a larger device. However, physics-based models indicate an important role for the seed island mechanisms, which may in fact give rise to increased stability on larger devices-i.e. if the seed island width (required to trigger the NTM) falls below the critical levels required. Fits based on these models suggest this is the case, bur are too badly constrained at present to make reliable predictions, and the physics is complex, making quantitative theoretical calculation difficult. Further experiments are required to examine the scaling of the seed, as well as to identify the role and relative sizes of the stabilizing terms that set the critical size for mode growth. In the event that the modes are unavoidable, promising feedback stabilization techniques are being developed with the use of localized RF current drive to change the stability properties of the plasma. Further work is needed to demonstrate sustained access to higher beta (N) and provide data to refine models. This paper reviews the underlying physics and key issues, commenting on the present status of understanding and further work required.