The theory of grain boundary migration as a thermally activated process is reviewed, the basic mechanisms in ceramics being the same as in metals. However, porosity and non-stochiometry in ceramic materials give an added dimension to the theory and make quantitative treatment of real systems rather complex. Grain growth is a result of several simultaneous (and sometimes interacting) processes; these are most easily discussed separately, but the overall rate depends on their interaction. Sufficient insight into the nature of rate controlling diffusion mechanisms is necessary before a qualitative understanding of boundary mobility can be developed.