Accurately simulating hydrological processes such as streamflow is important in land surface modeling because they can influence other land surface processes such as carbon cycle dynamics through various interaction pathways. This study aims to evaluate the global application of a recently developed Model for Scale Adaptive River Transport (MOSART) coupled with the Community Land Model (CLM4.0). To support the global implementation of MOSART, a comprehensive global hydrography dataset has been derived at multiple resolutions from different sources. The simulated runoff fields are first evaluated against the composite runoff map from Global Runoff Data Center (GRDC). The simulated streamflow is then shown to reproduce reasonably well the observed daily and monthly streamflow at over 1,600 world's major river stations in terms of annual, seasonal and daily flow statistics. The impacts of model structure complexity are evaluated and it is shown that the spatial and temporal variability of river velocity simulated by MOSART is important for capturing streamflow seasonality and annual maximum floods. Other possible sources of the simulation errors are linked to uncertainties in the forcing, as revealed by simulations driven by four different climate datasets; and human influences, based on a classification framework that quantifies the impact levels of large dams on the streamflow worldwide.