Agroforestry systems are generally characterized by several parameters, such as the number of plant species, the density of each species, or various biodiversity indexes. Many systems have a main crop, like coffee or cacao. In such systems, the ratio of the number of trees of the main crop to the total number of trees (the main crop plus the associated species) is also used to characterize the system. This parameter can vary from 1, in the case of a pure crop, to a value near 0 in more complex systems. This parameter assesses the relative density of the main species, but it does not provide an idea of the organization of a system. Indeed, for the same relative density, the system may be completely spatially regular or completely disordered with pockets of the pure main crop alternating with pockets without main crop trees. To estimate the level of organization in agroforestry systems with a main crop, we propose an index which varies from 0 to 1. For a given area, the plot is split into n sub-plots with the same area. In each sub-plot, the relative density is estimated, i.e. the number of trees of the main crop out of the total number of trees. The index En is expressed as the ratio of the relative minimum density obtained in one of the n sub-plots to the relative density estimated for the whole plot. In the case of a very regular system En =1 whereas in very irregular systems En -> 0. A second index of the organization of the system is the speed of convergence of En to E1 when n decreases (n -> n/2 ...-> 1). Several examples are presented to illustrate the relevance of these parameters, which could help to more effectively compare agroforestry systems. (Texte intégral)