The overall objective of this thesis is to understand the role of diversification at household level, its patterns, determinants and effects on rural smallholder farmers in SSA, as well as its implications for the process of structural transformation in SSA. In the thesis, I use a mixed methodology and combine different types of data on rural smallholders in SSA to answer the objectives a review of previous literature, empirical data from quantitative surveys and qualitative fieldwork. The findings are presented in three articles: In the first article, I undertake a comprehensive review of the literature on the nature and evolution of rural livelihood diversification in SSA among smallholder farmers. It reveals mixed findings about the causes and consequences of livelihood diversification on the rural smallholders adopting this strategy. It shows that because of asset constraints increase in incomes and wealth based on livelihood diversification has not yet benefitted the large majority of smallholders in SSA. On the other hand, there are a lot of evidence from the literature suggesting that it is relatively better-off smallholders with sufficient assets, who achieve successful livelihood diversification, mainly by exploiting opportunities and synergies between farm and nonfarm activities. In the second article, I investigate the spatial and geographical patterns and determinants of income diversification using cross-sectional data on rural farm households from six regions in Senegal and Kenya. In addition, I supplement the analysis of the quantitative data with information from qualitative fieldwork. The findings show that the specific patterns and determinants of diversification differ significantly between regions, with push and pull factors sometimes acting concurrently. Although geographical location matters for income diversification, the context of the region seems to matter even more, as it influences the type of diversification households may engage in. In the third article, I use panel data to explore the geographical and gender dimensions of livelihood diversification and its determinants in two agricultural regions of rural Kenya (Kakamega and Nyeri). I complement the panel data with data from my own qualitative fieldwork. The empirical investigation into the dynamics and motivations for livelihood diversification shows that household asset wealth, the initial level of diversification, demographic factors such as age, gender (being a female headed household) and level of education of the household head (in Kakamega), and increased use of hired labour were significantly positive determinants of change in livelihood diversification. Whereas, increased access to farm-related assets such as agricultural input credit, and having more secure land rights through formal titling, promoted specialisation in farming rather than diversification out of farming. Finally, food security was important for increasing livelihood diversification, especially in Kakamega.