LAND SCARCITY is a recurrent public concern with important policy implications. Growing demand -for food as well as clean energy, urbanization, ecosystem services or wildlife conservation- challenges the capacity of available resources to fulfill these needs. On the matter of land scarcity, often characterized in terms of available acreages (in hectare), the objective of this thesis is to define and study its qualitative dimension. Because land resources are heterogeneous and spatially fixed, the augmentation of supply - in particular by changing land uses - is constrained by the existing distribution of non-producible attributes (e.g. soil quality, climate, accessibility). This recognition is not new, (Ricardo, 1817), but we propose here some extensions. From a theoretical point of view, we develop a model of optimal land-use allocation with multidimensional heterogeneity (one for each use). It connects land use, land value and social preferences. In a competitive land market, the allocation is determined by comparative advantages. Following a shift in preferences, the law of diminishing returns of land is presented as a special case of a model that admits increasing returns. With market failures, we show the importance of information about the distribution of heterogeneity -marginal variations and the correlations between them- to implement public action. However, the model presents some situations when additional information does not imply the possibility to make better choices. In general, these results move the analysis of land scarcity in direction of statistically measuring the value of different uses of the same resource. From an empirical point of view, we study land heterogeneity by land price (Côte d’Or, 1993–2005) and by observed choices of farmers (bassin parisien, 1992–1993). The first work relies on observed prices, soil attributes (water-holding capacity, textures, hydromorphy), and topography (altitude, slope) within an hedonic framework. We obtain cumulative effects of these land characteristics which explain approximately 35–60% of land value. The second work consists in revealing the opportunity costs of cropland diversion by using 1992 CAP reform as a quasi-experiment. Under the compulsory set-asides required to regularly obtain direct payments, farmers’ choices correspond to benefit/cost trade-offs which, by observing benefits, permit the computation of the costs. Land heterogeneity is then identified by the spatial pattern of opportunity cost, after controlling for farm-level economic structure.The last part of the thesis uses land heterogeneity to study the equity of simulated biological conservation policies on farmland in Provence region. The administrative scale at which policy objectives are determined (région or departement) can increase or decrease local land scarcity. With regional objectives, conservation efforts are concentrated in administrative units that have better land endowments (Var and Vaucluse).With local objectives conservation efforts are concentrated in the administrative units with the poorest land endowments (Alpes-Maritimes). Equity considerations allow us to find an interior solution to this trade-off.