Agroforestry is promoted as a practice at the crossroads of sustainability and productivity objectives; however, many agroforestry programmes have had mixed effects due to a lack of understanding of the compatibility of the species supplied to farmers with cocoa and a failure to take account of their knowledge in designing the programmes. This paper, therefore, examines the effects of socio-economic and agroforestry factors on cocoa yields in Côte d’Ivoire, West Africa. The data used come from surveys of 150 farmers in three areas of the country: Bonon, Soubré and Biankouma. The choice of these areas was based on an east–west gradient, reflecting the evolution of the cocoa loop. The Bayesian Information Criterion method and multiple linear regression were applied to identify the species and their relationship with yield. The results showed that certain species, such as Citrus sp., Cordia senegalensis, Isoberlinia doka, Morinda lucida, Morus mesozygia and Raphia hookeri increased in yield; on the other hand, Anthonotha manii was found to reduce in yield. Finally, labour and insecticides contributed to yield increases. The statistical analysis can be supplemented with agronomic and ecological analyses to improve species management on cocoa farms.