An approach for estimation of the parameters of a macroscale land surface hydrology model is illustrated for the Global and Water Cycle Experiment (GEWEX) Continental Scale International Project (GCIP) large-scale area southwest (LSA-SW) which essentially comprises the Arkansas-Red River basin. The macroscale land surface hydrology model parameters were estimated for 44 catchments within LSA-SW with areas ranging from 180 to 7100 km2 using an automated search procedure. The catchment parameters were then linearly interpolated and overlaid on a one degree grid, which was used to represent the drainage network. The macroscale grid network model was run off-line at a daily time step, forced by gridded station precipitation and potential evapotranspiration. The model-generated long-term mean streamflows were compared with observations (corrected for management effects such as reservoir storage and diversions) and were found to agree to within one percent for the Arkansas River and about two percent for the Red River. For both rivers, the model underestimates the seasonal peak streamflow in late spring, and overestimates the late summer and early fall minimum. Model-derived evapotranspiration, spatially averaged over the entire Arkansas-Red basin, was compared to evapotranspiration derived from an atmospheric moisture budget of the Arkansas-Red River basin. On an average annual basis, for the period 1973--1986, the two agree to within one percent. The mean seasonal cycles for the two estimates agree quite closely from late winter to midsummer. However, the hydrologic model estimates less evapotranspiration in the fall, and more in midwinter, than the atmospheric budget. ¿ American Geophysical Union 1996