In recent decades, many parts of the African continent have experienced high precipitation variability with periodic drought and flood events. However, the network of streamflow gauges is too sparse in most countries to adequately capture these variations. In addition, no observed reference climatological dataset exists to adequately represent precipitation and temperature changes within all topographic and climatic zones. Consequently, the use of global gridded datasets needs to be considered. This paper aims to use the different available gridded datasets as inputs to a hydrological model to evaluate dataset performance. Nine precipitation and two temperature gridded datasets are used to this effect. The precipitation datasets include two gauged-only products, two satellite products corrected using ground-based observations, four reanalysis products and one merged product of gauge, satellite, and reanalysis.The two temperature datasets include one gauged-only and one reanalysis product. The ten precipitation and two temperature datasets were combined in their 18 possible arrangements for analysis purposes. Each combination was used to force the HMETS lumped hydrological model. The model parameters were calibrated individually for each combination against
the streamflow records of 850 African catchments. The Kling-Gupta Efficiency (KGE) was used to evaluate the simulation performance. Results show thatboth temperature datasets performed equally well. Large differences were however observed between precipitation datasets. The MSWEP mergedproduct was the best-performing precipitation dataset, followed by CHIRPS satellites and ERA5 reanalysis products, respectively. The performance of both gauged-only datasets (CPC and GPCC) was inferior, outlining the limitations of extrapolating information in data-sparse regions.