Considerable progress in continental hydrological research has been made recently that includes sub-grid-scale variability of precipitation, canopy interception, through flow, and hill-slope runoff production. Many such studies have focused on one single process at a time. This work aims to examine the interplay between canopy interception, throughflow and runoff using the NCAR Community Atmosphere Model (CAM2) coupled with NCAR Community Land Model (CLM2). The simulations are compared with the observed runoff from the University of New Hampshire-Global Runoff Data Center (UNH-GRDC) 0.5 degree monthly climatological composite runoff fields. The simulations from the CAM2/CLM2 show an excessive canopy interception and low soil moisture content over the climatologically wet regions, especially in the Amazon basin. A series of experiments are conducted which explicitly account for the sub-grid-scale variability of precipitation and its effects on canopy interception and throughflow. These experiments incorporate changes in the treatment of topography-induced runoff. The results have demonstrated that the interception and runoff processes are intimately coupled, and that their schemes should be changed together to ensure the improvement in the hydrological simulations. The simplified topography-based runoff model (TOPMODEL) and the sub-grid interception scheme proposed in this study, when used together, result in better simulations of runoff. With these proposed schemes, the water budgets show favorable changes in the Amazon basin in that canopy evaporation is reduced, runoff is increased, and soil is wetter.