In this paper, a model for hydrological equilibrium of leaf area index (LAT) is presented. The central assumption is that adaptation of vegetation to the local climate leads to a development of the canopy, which ensures a large light absorption but prevents severe soil water depletion. Particular attention has been paid to the processes which strengthen or weaken the basic LAI/evapotranspiration relationship, including soil moisture feedback on transpiration and evaporation, deciduousness of the canopy and interception losses. The annual water balance has been compared to stream flow data. A general agreement was found between simulated and observed run-off averaged over 28 large river basins. This shows that a water exchange scheme with explicit canopy processes can be used on a global scale. Vegetation aridity gradients are well defined by the hydrological equilibrium theory. An additional light availability constraint shows that the LAI, for wet tropics, some wet temperate and boreal areas, is not primarily water limited. An interesting point arises from an underestimation of the LAI and evapotranspiration in the rain forest margins and dry forests. This suggests that these areas are prone to recurrent droughts, and that drought resistance and avoidance strategies are of uppermost importance for vegetation function and therefore water cycle modelling. In particular, deep rooting may play a crucial role. The objective of modelling vegetation response to resources availability-here the water and light resources-is to improve our understanding of the biosphere as a resource-based system on a global scale, in the perspective of a change in atmospheric CO2 and corresponding climate modifications. The exact role of below-ground functioning, its links with carbon allocation, and especially the cost of deep rooting, emerge from this study as critical questions. (C) 1998 Elsevier Science B.V. All rights reserved.