Understanding the factors that determine the variability of intra- and inter-orchard production is crucial to help steer agronomic research and improve yield. Few studies have focused on identifying these factors in mango orchards in Côte d'Ivoire, one of the major mango producing countries of West Africa. This work quantifies mango production and analyses the effects of both tree dimensions and year on production levels. In 23 orchards, 230 mango trees ('Kent') were selected to be representative of tree dimensions and productivity in the Poro region of northern Côte d'Ivoire. Tree dimensions measured were tree height, trunk height, trunk circumference and crown dimensions. Individual tree production was estimated in 2019 and 2020 by image analysis. Before harvest, each tree was photographed on two opposite sides with a RGB camera. Images were analyzed with a machine learning algorithm combined with a linear model. Similar to other countries in West Africa, a high variability in production was observed between trees in 2019 and 2020. Trees were more productive in 2019 than in 2020. Mango production dropped in 2020 for about 75% of the trees, in particular large trees. In 2019, tree production was positively correlated with tree dimensions, whereas these correlations were weak and negative in 2020. Mango tree production in northern Côte d'Ivoire showed a high variability between trees and between years that could not be explained by tree structure alone. This high variability could be reduced by the adoption of cultural practices such as tree pruning, irrigation if rainfall is not sufficient, or fertilization if possible.