This discussion compares and contrasts parallel automated hydrologic forecast systems. Two forecast systems were run in parallel for the Skagit River watershed in Washington State to evaluate the effectiveness of incorporating meteorological observations in the reduction of error in peak streamflow forecasts. Both forecast systems employ guidance from a numerical weather forecast model (the Penn State/NCAR Mesoscale Model version 5, MM5) to drive a fully distributed hydrology model (the Distributed Hydrology Soil Vegetation Model, DHSVM). One system is driven with observation based initial conditions and the other is driven with MM5 based initial conditions.