Many instances of compensation behaviour (i.e. conformity of the Arrhenius parameters to a relationship of the form : lnA = bE(a) + c, where b and c are constants) reported for solid-state decompositions, refer to reactions that are at least partially reversible under reaction conditions used in the kinetic studies. Arrhenius parameters calculated for such processes are sensitive to the prevailing pressure of volatile product and heat transfer controls that may vary appreciably between successive experiments. Thus, compensation effects have been reported for various single reactions (e.g. the decomposition of CaCO3, the dehydroxylation of Ca(OH)(2) and the dehydration of Li2SO4 . H2O) where, for each solid, the reactivity of the starting material can be regarded as constant. Compensation has also been reported for sets of chemically comparable reactants that (are expected to) decompose in the same temperature interval. Compensation may then arise either from the aforementioned variation of reaction conditions, or from differences in the reactants, such as particle sizes, packing, etc. Because of the variety of compensation effects reported, the phenomenon is often regarded as an experimental artefact. It is of interest to examine the reasons for the concurrent changes of Arrhenius frequency factors and activation energies. The significance of activation energies in solid-state decompositions is discussed briefly and three classes can be distinguished. The accuracy of measurement of activation energies needs to be increased so that their sensitivity to prevailing reaction conditions can be established, investigated and understood.