AimCurrent weaknesses of diversity theory include: a failure to distinguish different biogeographical response variables under the general heading of diversity; and a general failure of ecological theory to deal adequately with geographical scale. Our aim is to articulate the case for a top‐down approach to theory building, in which scale is addressed explicitly and in which different response variables are clearly distinguished.LocationThe article draws upon both theoretical contributions and empirical analyses from all latitudes, focusing on terrestrial ecosystems and with some bias towards (woody) plants.MethodsWe review current diversity theory and terminology in relation to scale of applicability. As a starting point in developing a general theory, we take the issue of geographical gradients in species richness as a main theme and evaluate the extent to which commonly cited theories are likely to operate at scales from the macro down to the local.ResultsA degree of confusion surrounds the use of the terms alpha, beta and gamma diversity, and the terms local, landscape and macro‐scale are preferred here as a more intuitive framework. The distinction between inventory and differentiation diversity is highlighted as important as, in terms of scale of analysis, are the concepts of focus and extent. The importance of holding area constant in analysis is stressed, as is the notion that different environmental factors exhibit measurable heterogeneity at different scales. Evaluation of several of the most common diversity theories put forward for the grand clines in species richness, indicates that they can be collapsed to dynamic hypotheses based on climate or historical explanations. The importance of the many ecological/biological mechanisms that have been proposed is evident mainly at local scales of analysis, whilst at the macro‐scale they are dependent largely upon climatic controls for their operation. Local communities have often been found not to be saturated, i.e. to be non‐equilibrial. This is argued, perhaps counter‐intuitively, to be entirely compatible with the persistence through time of macro‐scale patterns of richness that are climatically determined. The review also incorporates recent developments in macroecology, Rapoport’s rule, trade‐offs, and the importance of isolation, landscape impedance and geometric constraints on richness (the mid‐domain effect) in generating richness patterns; highlighting those phenomena that are contributory to the first‐order climatic pattern, and those, such as the geometric constraints, that may confound or obscure these patterns.Main conclusionsA general theory of diversity must necessarily cover many disparate phenomena, at various scales of analysis, and cannot therefore be expressed in a simple formula, but individual elements of this general theory may be. In particular, it appears possible to capture in a dynamic climate‐based model and ‘capacity rule’, the form of the grand cline in richness of woody plants at the macro‐scale. This provides a starting point for a top‐down, global‐to‐local, macro‐to‐micro scale approach to modelling richness variations in a variety of taxa. Patterns in differentiation/endemicity, on the other hand, require more immediate attention to historical events, and to features of geography such as isolation. Thus, whilst we argue that there are basic physical principles and laws underlying certain diversity phenomena (e.g. macro‐scale richness gradients), a pluralistic body of theory is required that incorporates dynamic and historical explanation, and which bridges equilibrial and nonequilibrial concepts and ideas.