The Sahel, a water-vulnerable region in West Africa, relies heavily on rainfed agriculture. The region experienced pronounced droughts during the 20^th Century, emphasising the socio-economic importance of understanding the drivers of the rainfall variability. However, future rainfall projections remain uncertain due to the complex nature of the West African Monsoon (WAM), which is influenced by internal climate variability, external forcing, and feedback processes. Limited observational records in West Africa and the need for longer time series further complicate the understanding of these drivers.

 This thesis uses paleoclimate modelling to investigate internal and external drivers of monsoon variability in West Africa across four distinct periods. Our study confirms that atmosphere-only model simulations can capture the observed multidecadal rainfall variability in the 20^th Century, even though reanalyses struggle to reproduce the correct timing. Analysis of a last millennium simulation using the Earth System Model EC-Earth3 identified two drivers of multidecadal rainfall variability, accounting for 90% of the total co-variability between the West African rainfall and Atlantic sea surface temperatures (SSTs). This finding strengthens our understanding of SST-WAM relationships observed during the 20^th Century. An ensemble of climate model simulations (PlioMIP2) shows that high CO₂ levels and a different paleogeography during the mid-Pliocene Warm Period led to increased rainfall and a strengthened WAM. Our study emphasised vegetation's crucial role in enhancing the monsoon in past climates.

 However, simulations forced with prescribed vegetation only capture a one-directional forcing. A mid-Holocene simulation using an Earth System Model with dynamic vegetation revealed that vegetation feedbacks strengthen the WAM response to external orbital forcing but are insufficient to shift the monsoon northward or increase vegetation cover over the Sahara. These results reveal a dry bias and under-representation of simulated vegetation compared to proxy records, highlighting the importance of model development and the need for additional feedback processes in driving an enhanced, northward WAM and extending vegetation to the Sahara.

 Overall, this thesis advances our understanding of the drivers of West African monsoon variability and provides valuable insights for improving future rainfall projections in this vulnerable region.