The binary exchange of cations on clays and soils is generally regarded as a thermodynamically reversible process. The literature on soil chemistry and geochemistry, however, abounds with reports on cation exchange reactions that appear to have only limited reversibility, i.e., that exhibit hysteresis. A satisfactory explanation of this phenomenon is still lacking, even though a number of mechanisms have been advocated, e.g., charge or site heterogeneity at the surface, differential hydration of cations, dehydration of the exchanger, crystalline swelling hysteresis, and inaccessibility of sites caused by domain or quasi-crystal formation. In the present article, the relevant literature is reviewed and analyzed critically. On the basis of available evidence, it is shown that exchangeable cations can be classified into three groups, defined in such a way that hysteresis has, in the literature, generally not been observed when exchange reactions involved cations belonging to the same group, but has often been found when the reactions involved cations from different groups. Furthermore, it is argued that none of the five mechanisms mentioned can, in and of itself, account fully for the observed exchange hysteresis. A conceptual model is proposed that combines elements of these five mechanisms and is able to describe, at least qualitatively, the effects of factors such as clay type, electrolyte concentration, and extent of dehydration.