A central goal of ecological studies is to understand the behaviour of species in their natural environments. A corollary of this is that if we understand that behaviour, we should be able to predict what happens. A methodology then emerges, beginning with the description of processes, leading to the development of models and ending with the testing of their predictions. The study of plant population biology has, for the most part, followed this pattern. There have been many studies of monocultures and simple mixtures in pots, often in the glasshouse. A considerable volume of papers has used mathematical models to investigate what, in theory, could happen to populations of both plants and animals as a result of intrinsic processes. However, attempts to test these predictions on plant populations in the field are hard to find (e.g. Symonides et al. 1986) and even then they may be equivocal (Rees & Crawley 1991). Why is this? Has plant population biology lost its way? Are few researchers trying to bridge the gap between theory and reality? Or are there practical problems which are insurmountable? The environment of a plant population is variable and, in many respects, unpredictable. Long-term monitoring studies may be impressive and appear to be valuable sources of information (e.g. Inghe & Tamm 1985) but the nature of the intrinsic population dynamics is expected to be masked by considerable environmental 'noise', by the complexity of the habitat/community and by sampling error. This difficulty with detection has even led to the conclusion that density-dependence does not occur in many communities. Many plant population biologists, however, choose to work over a short period, usually of less than one generation. Experiments can be designed to minimize environmental noise; density dependence is then invariably shown. Through models based on these data, complex patterns of behaviour have been deduced as being possible in theory and, there, many studies end. The ultimate objectives of most researchers must usually be inferred from the aims of various isolated papers which cover only part of their research area. Perhaps these objectives are, or have become, restricted to simply generating short-term data and predicting the resulting long-term dynamics in the (artificial) absence of environmental noise? Perhaps this is sufficient to complete one piece of our ecological jig-saw but it is surely not sufficient as the objective of plant population biology in its entirety? It is clear that at least some researchers have more optimistic expectations of population models. For example, Thrall et al. (1989) state that, 'In general, statistical analyses of time-series of census data will not necessarily allow one to determine whether fluctuations in population size are caused by densitydependent factors or are environmentally driven noise. Perhaps the only reliable way to demonstrate that oscillations are caused by density-dependent interactions is to calibrate a quantitative model of density-dependent regulation at a single point in time.' Begon & Mortimer (1986) consider that, because a deterministic mathematical model predicts that density-dependence can generate population fluctuations, '... we need not look beyond the intrinsic dynamics of a species in order to understand the fluctuations in numbers of its natural populations'. Not only are these expectations optimistic but they are also philosophically untenable. Even if a realistic model can be formulated, its predictions can only tell us what might happen in a constant environment. Unvalidated deterministic density-dependent models based on short-term data are not sufficient to confirm cause and effect of long-term dynamics in the field. Plant population biology must surely be concerned with the actual and not just the theoretically possible. How can we, for example, distinguish in the field between environmentally driven fluctuations of an otherwise asymptotic trajectory, pure chaos driven by density-dependence and stable cycles perturbed by environmental noise? The aim of this paper is to discuss briefly some of the current methodologies used to study the effects of intrinsic factors on plant population dynamics and to examine the extent to which they are bridging the gap between theory and reality.