Plant movements have fascinated scientists since at least the end of the 19th century. The occurrence of these movements rises many questions about the way they are regulated by the plant (mechano-perception, signal transduction and trigger of the movement) as well as about the nature of the “motor”, namely the mechanisms that provide the mechanical energy for these movements. Rapid movements (such as e.g. seed ejections or carnivorous plants) are often based on the accumulation of elastic energy, either by osmotic pressure or by desiccation. The motor of slower movements of plant young stems and leaves is also well-identified, based on changes in cell turgor pressure inducing differential growth and thus bending movements. However, for stiffer organs such as lignified stems, the magnitude of the mechanical stress provided by these turgor-related mechanisms is not enough to generate significant bending movements. The motor forces are provided by the so-called “maturation stress”, a mechanical stress that is set-up in lignified cells during the formations of secondary cell-walls. Occurrence and magnitude of these maturation stresses (up to tens of MPa) in wood have long been evidenced experimentally. Their consequences on the internal state of stress of tree trunks and the way they induce stems movements are well understood and consistently explained by biomechanical models. Nevertheless, at sub-cellular scales, the mechanism by which motor forces are generated during cell-wall formation is still a matter of debate and an active research topic. Woods generating mechanical stresses of large magnitude are called reactions woods. They are classically divided into “compression woods” that generate compressive stress on the lower side of tilted stems of gymnosperms, characterized by large microfibril angle (MFA) and high lignin content, and “tension woods”, generating tensile stress in the upper side of tilted stems of angiosperms, and typically characterized by a specific “gelatinous” layer, devoid of lignin and having a very low MFA. Many works have evidenced that this classification is a simplification, some examples not conforming to it, such as compression wood found in some angiosperm species, tension wood found in gymnosperms, and angiosperm tension woods in which the G-layer is not apparent. Nevertheless, we will here concentrate of the mechanisms involved in typical compression wood and G-layer tension wood. Comprehension of how compression wood works resulted from the analysis of structure /properties relationships, namely the link between MFA and the magnitude of maturation strains. By analogy with previous modeling of the drying shrinkage of compression wood, researchers have assumed that the source of mechanical energy was a dimensional variation (here swelling) of the lignified matrix, and that this movement was directed across the stiff microfibrils, so that, if the MFA is large, the stress is directed in the direction of the cell axis. In order to model jointly the maturation stress of both normal and compression wood as a continuum, it was necessary to assume that a second process was also at work, namely a shrinkage of the cellulose network along the microfibril direction. This model does account correctly for both the sign and order of magnitude of the maturation stress, in both the longitudinal and tangential directions of woods, for both normal and compression woods. However, it is still limited in at least two aspects: (1) it provides an explanation for the role of lignification and MFA in the process, but does not identify at the molecular level what causes the swelling of the matrix and the shrinkage of the cellulose network; (2) it is not sufficient to explain the large tensile stress found in tension wood. Regarding tension wood, a number of new experimental observations have been reported in the recent years, and various models have been proposed to explain the link between its structure and its biomechanical function. Most of these models are based on a description of the interaction between the unlignified matrix and cellulosic microfibrils. In this presentation, I will ty to summarize the models that have been proposed and confront them to the body of knowledge that has been accumulated in the literature. I will argue for a possible model based on the swelling of the matrix in a connected microfibril network. Although a lot of progress has been achieved in the biochemistry and molecular biology of the formation of the G-layer, the mechanism that induces the swelling of its gel-like matrix material is still not fully understood. Recent progress on the study G-layers located in other tissues than wood (e.g. cortical fibers of annual plants) and on tension woods without visible G-layer may contribute to build soon a consistent picture of the mechanisms generating in high magnitude stress in plants, encompassing normal woods and all kinds of reaction woods.