This paper provides a comparison of two widely used global precipitation datasets in southern Africa: Tropical Rainfall Measuring Mission (TRMM) and Water and Global Change (WATCH) Forcing Data (WFD). We also evaluate the performance of the water and snow balance modelling system (WASMOD-D) in a water balance simulation of 22 gauged basins over the southern Africa region. Water balance for southern Africa was simulated using the two global datasets as input with regionalized model parameter values. The results reveal that the special variation patterns of mean annual precipitation from TRMM and WFD datasets and temporal changing trend are similar in southern Africa. However, they are quite different in terms of probability distributions. Simulation of WASMOD-D based on two datasets in southern Africa results in model performances of above 0.5 for Nash–Sutcliffe (NS) values, below 10% for volume error (VE) values and a good reproduction of the observed flow duration curves for the majority of the basins. Finally, WFD data which have been bias corrected were observed to outperform TRMM data in southern Africa. The approaches and results described in this study contribute to the limited literature on regional hydrological modelling in the data-scarce region of southern Africa.