The multi-agency/university North American Land Data Assimilation System (N-LDAS) project is designed to provide enhanced soil and temperature initial conditions for numerical weather/climate prediction models by using real-time observed precipitation and solar insolation data. Currently four different land surface models (LSMs) are running in N-LDAS in both retrospective mode as well as in realtime. All LSMs are initiated at the same time with the same relative soil wetness. This study examines the degree of correlation between the water balance simulation and how this may vary with time. The mean statistical properties as well as the spatial variation of the water balance components, including runoff, evapotranspiration and soil moisture, are investigated. The results of this study should provide important insights into the similarities and differences of the four LSMs in N-LDAS. Further, this study should also shed light on the spin-up properties and possibility of using soil moisture states from one model to estimate initial soil moisture states from another model.