Distributed meteorological data collected from different sources are rarely identical within the same domain of space and time. Discrepancies of these data in magnitude, pattern, and resolution play an important role in hydrological simulation. Using four different sets of distributed meteorological data (from the HUBEX-Intense Observation Period and GAME experimental products at different resolutions), hydrological simulations are conducted through a distributed hydrological model called MaScOD (macro-scale OHyMoS assisted distributed) hydrological model. The model's performance is measured using 12 different indexes. Based on these indexes, a relative normalized score is calculated to evaluate the overall performance of the result from each data set. Three sub-basins of the Huaihe River basin in China, taking the cases at Bengbu (132 350 km(2)), Wangjiaba (29 844 km(2)) and Sniping (2093 km(2)), are used for numerical experiments. This study shows the competence of coarse-resolution meteorological data, the GAME reanalysis 1(.)25degrees data, to apply in hydrological simulations of large catchments. However, that data failed to simulate the hydrograph in smaller catchments. The results are significantly improved by including spatial variability at finer resolution in that data. Copyright (C) 2004 John Wiley Sons, Ltd.