The routing of runoff estimates from General Circulation Models (GCMs) is i
mportant to model river flow from large river basins and to estimate freshw
ater inflow into the oceans. Present routing approaches use arbitrary const
ants and empirical equations to determine travel times between the GCM grid
cells. A new river flow routing scheme is developed, which uses physical c
atchment and river channel information and river discharge data. The scheme
uses surface runoff and deep percolation data from land surface parameteri
zation schemes, and routing is performed via linear surface and groundwater
reservoirs. Geomorphological relationships between mean annual river disch
arge and other physical variables are used to interpolate existing informat
ion to define the channel morphology for the digital river networks at the
routing model resolution. Applications of the routing scheme to the Mississ
ippi and Amazon River Basins indicate that it performs adequately. The sche
me does not require the calibration of parameters and can thus be easily us
ed in GCMs.