Improvements to traditional flood forecasting can be made through the integration of short-term radar nowcasting and distributed hydrologic modeling. For short-lead times (0-3 hrs), the extrapolation of NEXRAD-based rainfall estimates can improve the lead time and accuracy available for issuing flood forecasts as compared to rainfall persistence or climatology. Distributed models explicitly account for rainfall variability and provide forecasts of the lumped and distributed basin response after appropriate calibration. In this paper, we present the use of the MIT Lincoln Lab Growth and Decay Stormtracker (GDST) as a quantitative precipitation forecast (QPF) input to the MIT TIN-based Real-time Integrated Basin Simulator (tRIBS). Flood forecasts issued by the distributed model based on the extrapolation QPF are developed for two adjacent watersheds within the Arkansas Red River Basin for selected storms within the 1998-2001 period. Comparisons to observed streamflows at the watershed outlet and interior gauging stations demonstrate the utility of the nowcasting product. An analysis of the error propagation from the extrapolation product through the flood forecast is also presented as a function of forecast lead time. Finally, the impact of storm, basin and radar rainfall scales for the case studies is discussed.