Soil fertility decline is a reality across the globe. Soil infertility has led to falling crop yields for both food and cash crops precipitating both poverty and food insecurity. Poor agricultural practices have often been blamed for declining levels of soil fertility. It was within this framework that this study was carried out to assess soil fertility levels in cocoa-based agroforestry systems in the Littoral region of Cameroon. Through the use of a mixed research approach and descriptive and inferential statistical analysis it was found that the most perceived indicators of soil fertility by cocoa farmers in cocoa-based agroforestry systems were cocoa yields (100%), soil colour (90%), indicator plant species (75%), presence of soil macro-organisms (80%), indicator weed species (65%), growth rate/vigour of the cocoa plant (70%), and level of compaction of the soil (50%). Chi-square test statistic (X2 = 15.92, p<0.05) revealed that there was a statistically significant difference in cocoa farmers’ perceptions of soil fertility indicators in cocoa-based agroforestry systems. Most cocoa farmers perceived the level of soil fertility in cocoa-based agroforestry systems to be between average (29%) and low (43%). Very few cocoa farmers perceived that soil fertility level in cocoa-based agroforestry systems was very high (5.3%). Chi-square test statistic (X2 = 38.42, p<0.05), revealed the existence of a statistically significant difference in cocoa farmers’ perception of the level of soil fertility in cocoa-based agroforestry systems. Correlation and regression analyses showed the existence of a direct and inverse non-causal and causal relationship respectively between several explanatory variables and soil fertility in cocoa-based agroforestry systems. It was noticed that explanatory variables such as fertilizer/manure application, diversity of tree species in the cocoa agroforests, diversity of soil macro-fauna in the cocoa agroforests, and density of soil macro-fauna in the cocoa agroforests, all had a statistically significant (p<0.05) direct non-causal and causal relationship with soil fertility in cocoa-based agroforestry systems. Meanwhile, explanatory variables such as age of the cocoa farm, density of cocoa plants in the cocoa agroforests, and application of agrochemicals in the cocoa agroforests, all had a statistically significant (p<0.05) non-causal and causal relationship with soil fertility in cocoa-based agroforestry systems. It is therefore recommended that there should be more organic fertilizer/manure application, more diversity of tree species in the cocoa agroforests, and measures to ensure the diversity and density of soil macro-fauna in cocoa-based agroforestry systems.