Bentonites are proposed to be used as buffers in high-level radioactive waste repositories. The elevated temperatures in repositories may, however, affect bentonites’ desired properties. For instance, heating under dry conditions can cause cation fixation, potentially affecting swelling properties. The kinetics of mineral dissolution and precipitation reactions will equally be influenced by temperature. Redistributions of Ca-sulphates and -carbonates have been observed, as well as illitization of smectite. Illitization, however, has only been observed in laboratory experiments at large solution/solid ratios, whereas it has not yet been unambiguously identified in large-scale experiments. In many large-scale tests, cation exchange is the first observable geochemical reaction. In addition, an enrichment of Mg close to the heater is found in many such tests. The thermal gradient and (incongruent) smectite dissolution are suspected to play a role with respect to the Mg enrichment, but the underlying mechanism has not been unravelled so far. To predict the long-term performance of a bentonite buffer, numerical modelling is required in order to be able to simulate the reactions of all accompanying mineral phases. Smectites, which dominate the bentonite composition, are therefore particularly difficult to characterise, as their dissolution is often observed to be non-stoichiometric. Various model approaches exist to simulate smectite reactions, mostly based on kinetic rate reactions, ideally considering the effect of pH (congruent or incongruent dissolution), temperature and the degree of saturation of the solution. Reassessing and improving the thermodynamic/kinetic data of smectites are prerequisites for improving long-term buffer performance assessment.