This Thesis presents an approach for the study of plant water balance during drought stress, using a combination of in vivo NMR experiments and computer simulations. The ultimate aim is the interpretation of the NMR parameters in terms of physiologically relevant characteristics, such as cell dimensions and membrane permeability. Especially the latter has raised a growing interest in plant science, and up to now the measurement of this parameter in vivo was limited to single cells and short experiment time spans.NMR microscopy of plants yields information on various levels of organisation. The NMR images provide clear anatomical details, which have been used to monitor the response of stem growth rates to osmotic stress. On the tissue and cell levels, the NMR parameters T 2 (transverse spin relaxation time) and D app (apparent diffusion coefficient) provide information on the physical and chemical properties. Correct quantitative values for T 2 and D app are crucial for a useful interpretation. Therefore, Chapter 2 evaluates the accuracy of different fitting procedures.The physical and chemical properties can vary considerably between and within different tissues, cells, and intracellular compartments, resulting in distinctly different relaxation and diffusion characteristics for these compartments. The interpretation of these parameters is not straightforward. A numerical model of restricted diffusion and relaxation behaviour was therefore developed, based on Fick's second law of diffusion (Chapter 3). This model expands previous one-dimensional models to a two-dimensional space, consisting of multiple concentric cylindrical compartments, separated by membranes. Numerical simulation experiments using this model demonstrate the importance of modelling two-dimensional diffusion in relation to the effects of spatial restrictions, and spin exchange between the different compartments.This model has been applied to investigate the effects of diffusive exchange on the transverse spin relaxation times, the apparent diffusion coefficients, and the NMR signal amplitudes of water in plant cells (Chapter 4). For different multi-compartment model systems a Pulsed Field Gradient Multiple Spin Echo (PFG-MSE) experiment was simulated, and intrinsic physiological parameters, i.e. the bulk diffusion constant, the cell radius and the membrane permeability were afterwards extracted using common theoretical models. The results justify the use of these models to interpret the in vivo experiments, since meaningful diffusion constants, cell radii and membrane permeabilities can be extracted for a large range of conditions. This is still true if not all conditions of the theory are known or met, e.g . for intact plants.Chapters 5 and 6 study the effect of mild osmotic stress on maize and pearl millet by in vivo1H NMR microscopy, and water uptake measurements. Single NMR parameter images of (i) the water content, (ii) the transverse relaxation time ( T 2 ) and (iii) the apparent diffusion coefficient ( D app ) were used to follow the water status of the stem apical region during osmotic stress. The results are interpreted using the multi-compartment model (Chapter 4), tailored to suit plant cells. For this particular case, an equation was derived to describe the relation between the observed T 2 , the cell dimensions, the bulk T 2 , and the membrane permeability, based on the Brownstein & Tarr theory. Experimentally determined T 2 values of non-stressed stem tissue are indeed correlated to the cell dimensions, which is in agreement with the derived equation. The T 2 of maize cells is higher than the T 2 of equally sized millet cells, implying that the membrane permeability of the latter is higher.The growth rate was strongly inhibited by mild stress in both species, even though the water uptake was only mildly affected. During stress, there are hardly any changes in water content or T 2 of the stem region of maize. In contrast, the apical tissue of pearl millet showed a ~30% decrease of T 2 within 48 hours of stress, whereas the water content and D app hardly changed. This decrease in T 2 can be caused by a decreasing cell radius, a decreasing bulk T 2 , and/or an increasing membrane permeability for water. To distinguish between these contributions, additional scanning electron microscopy was used, showing no apparent changes in cell size. Transverse spin relaxation measurements of a wide range of sugar solutions showed only very small effects of osmotic adjustment on the bulk T 2 . Together, these results point to an increase in membrane permeability during stress. This conclusion is confirmed by numerical simulations of the plant cell model, which showed that only an increasing membrane permeability yields a similar combination of water content, T 2 , and D app values during stress.Under severe osmotic stress, the effects on the plant water balance are naturally larger (Chapter 7). During stress, no significant changes occurred in the maize stem, though the leaves wilted, and the plant died after two days of stress. Pearl millet showed again changes in T 2 , especially in the secondary shoots, which were more pronounced than during mild stress. Furthermore, the stem tissue shrunk, implying that the cell dimensions changed; the secondary shoots showed far less decrease in water content, however. Despite these changes, the plants recovered once stress was relieved. In the framework of the plant cell model, the decreasing T 2 is interpreted as the result of a combination of decreasing cell size and increasing membrane permeability. The latter can result in a higher tissue conductance, thereby facilitating water re-allocation to young, expanding tissues to prevent irreparable damage.The combination of experimental data and simulations as presented in this Thesis has proven to be an effective tool to link NMR information to physiology (Chapter 8). This approach promises to be of great use to plant science, and to NMR microscopy in general.