Mango is a tropical crop characterized by the heterogeneity in size and quality of its fruits at harvest. Source-sink relationships have been identified as a key factor on these traits. Consequently, it is of main importance to characterize foliage distribution of the mango tree over time to assess its ability to produce fruits. Mango tree is an evergreen tree and its leaves last for several years. While leaf appearance has been well characterized, leaf mortality has been poorly studied for now. The goal of this work was to characterize leaf mortality in the mango tree, and in particular the influence of age and light availability on this phenomenon. We digitized the 3D architecture of a 13-year-old mango tree with foliage information. Using this 3D mockup, we characterized the number of leaves per growth units at different depths in the structure and related growth units depth with their age. Using a model of light interception, we then estimated the amount of light intercepted by the different growth units of the tree and tested its influence on their number of leaves. Leaves were present up to the fifth growth unit from the axis end. Leaf mortality drastically increased with growth unit depth. For each level of depth, light environment had a significant influence on the number of leaves remaining on the growth unit. Leaves in unfavourable light conditions had a 2-time higher probability to fall compared to the leaves in the most favourable light conditions. The resulting leaf mortality probability will be included into a functional-structural plant model (FSPM) to simulate faithfully foliage distribution within the tree. Using this FSPM that simulates fruit growth, it will be possible to study the impact of the foliage distribution on the heterogeneity of fruit production and quality.