Eastern Boundary upwelling systems, such as the Benguela upwelling system, are the regions of largest warm biases in the global coupled climate models. A better understanding of the regional processes, in particular those related to air-sea interactions, is therefore a challenging issue. This study focuses on the observed subseasonal variability in the Benguela upwelling system based on satellite data over the 1999-2009 period. First, we document spatial patterns and time-frequency characteristics of the subseasonal SST variability using Extended Orthogonal Function and wavelet tools. This analysis allows identifying two dominant regimes of variability that explain 25% and 20%, respectively, of the total subseasonal SST variability along the coast: a submonthly regime with a dominant 11-days oscillation and a lower frequency intraseasonal regime with a dominant 61-days oscillation. Both regimes are consistent with Ekman transport dynamics and are modulated, to a large extent, by the local surface winds. They are characterized by a strong seasonal dependence that can be explained by the marked seasonality of the subseasonal wind stress activity and by the vertical stratification within the ocean surface layer. The large-scale atmospheric patterns associated with the two regimes of wind-driven SST variability are further examined. Results indicate an important role of the eastward propagating disturbances at the mid-latitudes associated with a wave number 4 pattern and suggest a role of the intraseasonal component of the Antarctic oscillation in the subseasonal variability of the Benguela upwelling system. The impact of the remote equatorial Kelvin wave propagating along the coast on the SST variability is discussed.