Evapotranspiration constitutes a vital component of the hydrological cycle and should be considered when making land-use decisions, particularly in water-scarce regions. Quantifying evapotranspiration spatially for resource-constrained regions in the Global South which have limited monitoring data, would be highly beneficial for informing water resource management. Freely available satellite-derived evapotranspiration products offer an opportunity to achieve this. This study evaluated the performance of a series of freely available satellite-derived evapotranspiration products namely, TerraClimate, NASA's PT-JPL, WaPOR, SMAP, GLDAS, FLDAS, SSEBop, and MOD16. At monthly time steps, evapotranspiration estimates from these products were evaluated against in situ evapotranspiration estimates from 14 flux tower stations across several different bioclimatic regions in the Global South: South Africa. The top five satellite-derived evapotranspiration products performed well (rs: 0.75-0.83), with some performing better in certain bioclimates, but WaPOR and FLDAS emerged as superior in many cases. An ensemble was produced by taking the mean and variability of the top five best performing products, and this matched or outperformed the best-performing individual products overall, achieving an average rs of 0.85, a Kling-Gupta Efficiency of 0.67, and a Percent Bias of -15.47 %. The ensemble also performed satisfactorily across all evaluated bioclimates and can provide evapotranspiration estimates for the region at relatively fine spatial (100 m) and temporal (one month) scales. These results and this ensemble model of satellite-derived evapotranspiration can now be further explored to better understand relative water-use of different land-use types and improve water resource management.