Assessment of four alternate flash flood warning procedures was conducted to ascertain their suitability for forecast operations using radar-rainfall imagery. The procedures include (1) areal mean basin effective rainfall, (2) unit hydrography, (3) time-area, and (4) 2-D numerical modeling. The Buffalo Creek flash flood of July 12, 1996, was used as a case study for application of each of the procedures. A significant feature of the Buffalo Creek event was a forest fire that occurred a few months before the flood and significantly affected watershed runoff characteristics. Objectives were to assess the applicability of the procedures for watersheds having spatial and temporal scale similarities to Buffalo Creek, to compare their technical characteristics, and to consider forecaster usability. Geographic information system techniques for hydrologic database development and flash flood potential computations are illustrated. Generalizations of the case study results are offered relative to their suitability for flash flood forecasting operations. Although all four methods have relative advantages, their application to the Buffalo Creek event resulted in mixed performance. Failure of any method was due primarily to uncertainties of the land surface response (i.e., burn area imperviousness). Results underscore the need for model calibration; a difficult requirement for real-time forecasting. ¿ 2000 American Geophysical Union