Models are simplified representations of a system, i.e. a limited part of reality. Structure and properties of specific models are chosen depending on the purpose they serve. These purposes include: summarizing data, assisting in the analysis of experimental data, testing of hypotheses, extrapolation of system behaviour beyond the conditions that were covered experimentally, decision-support in practice. Over the last decade we have been involved in the development of functional-structural plant models (FSPM) in the domain of plant production. FSPM treat plants as a proliferation of elementary units explicitly describing 3D plant structure, and include physiological processes (e.g., photosynthesis and/or transport of substances through the plant structure). The main systems of study were tillering in wheat (an annual field crop) and flower cane production in glasshouse-grown cut roses (an intensively manipulated perennial crop system). Using wheat as an example, we shall outline the different steps to construct and parameterize an FSPM. Even though, in the first instance, such models remain quite descriptive, there is a range of applications of models that generate a realistic representation of the 3D plant structure (e.g., light distribution in relation to structural properties, remote-sensing research, pest and disease dynamics in relation to structure). In the wheat study it was tested whether the cessation of tillering, i.e. the arrest of 'bud break', could be explained from changes in light absorption (quantity) and signal perception (red/far-red ratio). This study was only a first step towards integrating knowledge on how light signals affect the structural development of a collection of plants. Subsequent steps taken include modelling the hormonal network regulating branching as modulated by the red/far-red ratio. Some issues that need to be addressed in order to make further progress will be discussed. Rose growers try to manage and manipulate their crops such as to maintain the production of a high number of flower canes of a particular quality (i.e. essentially a mixture of morphological, biometrical and flower yield-related traits) over a prolonged period of time. The number of canes produced depends on the number of buds that break. Understanding bud break is a central issue for FSPM because quiescence or breaking of buds directly affects the overall 3D structure of the crop, as well as the local light climate. Some data on bud break will be presented and the question will be raised on how to design and parameterize a 'decision tree', adequately describing bud fate over time in relation to the position of the bud in the structure and its environment (e.g. degree of illumination). Similar questions pertain to modelling signal transduction and to modelling of source-sink interaction and the associated flows of carbon in the 3D structure. Modelling structure and volume poses specific questions like: to which extent are growth in length and volume coupled with dry matter allocation? In other words: what are the temporal dynamics of growth in weight and volume and to what extent are these synchronized, and modified by internal and external factors? Results from detailed measurements on rose internodes show that the link between internode volume and fresh weight is very tight and largely independent from growth temperature, developmental age and phytomer rank. Properties of the shoot that grow from a broken bud may be predetermined from bud development. Stresses a plant experiences result in changed structural properties (e.g. leaf size) long after recovery from stress. Apparently, mechanisms exist of 'predetermination' of properties of organs. These are issues that need to be explored further to make FSPM able to cope with changing environments.