In attempts to postulate dehydroxylation mechanisms for the clay minerals, three assumptions have generally been made: that the elements of water are lost more or less uniformly from all over the crystal, that the tetrahedral parts of the structure are less likely to be disturbed than are the octahedral parts, and that silica-rich material often presumed to occur as a secondary product is in fact pure SiO2. It is shown that all of these assumptions are doubtful. Dehydroxylation mechanisms in which the first assumption holds good are described as homogeneous, and it is shown that certain calcium minerals undergo dehydroxylation in this way. In contrast, where magnesium or other cations of similar size are predominant, dehydroxylation occurs by an inhomogeneous mechanism, in which no oxygen is lost from those parts of the crystal where oriented conversion to a crystalline product occurs; the oxygen for the expelled water comes entirely from other regions of the crystal which are wholly or partly converted into pores, while various migrations of cations occur. The dehydroxylation behaviour of serpentine, talc, kaolinite and pyrophyllite is briefly reviewed. A new approach to the dehydroxylation of clay minerals is suggested, based on the concept of inhomogeneous mechanisms.