The lack of observed data for the hydrological modelling of catchments across borders has hindered the management of transboundary water resources. This study investigated the implications of different degrees of hybridization of observed rainfall data using ERA5-Land reanalysis precipitation data in simulating streamflow in the Ruvubu River catchment across the Burundi-Tanzania border. The hydrological model of the Ruvubu River catchment was set up and iteratively updated using 0%–100% hybrid rainfall data, parameterized, simulated, and then evaluated against observed streamflow data at the catchment outlet. The findings show that the performance of the hydrological model decreased as the degree of rainfall data hybridization increased. However, model parameters satisfactorily compensated for input uncertainty when the model had hybrid rainfall data not exceeding 15% hybridization. Subsequently, errors in the simulated flow were also minimal. This implies that the simulated flow from the model when it has hybrid rainfall data not exceeding 15% hybridization can represent the simulated flow when it has 0% hybrid rainfall data. These findings show the need to develop thresholds of rainfall data hybridization for different global precipitation datasets in data-scarce transboundary river catchments.