Soil loss, fluvial erosion, and sedimentation are major problems in semi-arid environments due to the high associated costs of decreasing services such as provisioning and regulating water resources. The objective of this research is to analyse sediment yield in a mountainous semi-arid basin, paying special attention to the sources of sediment, the associated uncertainties, and the transport processes involved. The segregation hypothesis along a reservoir of the sediment coming from hillslopes or fluvial systems is also evaluated. For this purpose, bottom-set and deltaic deposits of a reservoir (110 hm3) in southern Spain have been measured and compared with basin erosion and fluvial transport monitoring over a 12-year period. The volume of sediment stored at the bottomset of the reservoir shows a relative match with parametric predictions based on the Revised Universal Soil Loss Equation/Modified Universal Soil Loss Equation hillslope models and rating curves, estimated as being between 7 and 13 t·ha−1·year−1. Similarly, the measured volume of deltaic deposit fits the average value of stochastic simulations from different bedload transport equations. These contributions represent 50–65% of the total volume measured regarding suspended sediment inputs, way above that described in previous works. This highlights the importance of considering bedload when estimating the useful life of reservoirs in semi-arid environments. The major differences in sediment grain size between hillslopes and river systems, and the size fractions measured along the reservoir, support the assertion of segregation hypothesis. Nonetheless, based on the processes observed and the uncertainty related to modelling, that assertion has to be taken with caution. At basin scale, a specific sediment yield of between 19 and 24 t·ha−1·year−1 has been estimated, which includes hillslopes and fluvial contributions. This rate is in the range of sediment yield reported for Mediterranean mountain areas of a torrential character. The pulse-like nature of the system and the spatial heterogeneity of fluvial and hillslope erosion rates points out the importance of considering mid to long-term and process-based approaches and emphasizes the limitations of annual estimations for management purposes.