© 2019 Elsevier B.V.We are experiencing a proliferation of satellite derived precipitation datasets. Advantages and limitations of their promising application in hydrological modelling application have been broadly investigated. However, most studies have analysed only the performance of one or few datasets, were limited to selected small-scale case studies or used lumped models when investigating large-scale basins. In this study, we compared the performance of 18 different precipitation datasets when used as main forcing in a grid-based distributed hydrological model to assess streamflow in medium to large-scale river basins. These datasets are classified as Uncorrected Satellites (Class 1), Corrected Satellites (Class 2) and Reanalysis – Gauges based datasets (Class 3). To provide a broad-based analysis, 8 large-scale river basins (Amazon, Brahmaputra, Congo, Danube, Godavari, Mississippi, Rhine and Volga) having different sizes, hydrometeorological characteristics, and human influence were selected. The distributed hydrological model was recalibrated for each precipitation dataset individually. We found that there is not a unique best performing precipitation dataset for all basins and that results are very sensitive to the basin characteristics. However, a few datasets persistently outperform the others: SM2RAIN-ASCAT for Class 1, CHIRPS V2.0, MSWEP V2.1, and CMORPH-CRTV1.0 for Class 2, GPCC and WFEDEI GPCC for Class 3. Surprisingly, precipitation datasets showing the highest model accuracy at basin outlets do not show the same high performance in internal locations, supporting the use of distributed modelling approach rather than lumped.