A simple model of light interception was designed to represent light resource variability in multi-strata multi-species agroforests. Dealing with many different species makes it unpractical to develop a detailed model of light interception taking into account structural characteristics of each species (foliage lay-out, distribution of leaf orientation, radiation characteristics of leaves, etc.) as it has been done in other cases (Cescatti 1997; Bartelink 1998). The hypothesis to be tested was that a simple 3D description of trees inside a plot with average crown “density” estimated by photographs would allow a reasonably precise description of light received at any point. The model was tested against hemispherical photographs taken at ground level