Remotely sensed precipitation estimates and regional atmospheric model precipitation forecasts provide rainfall data at high spatial and temporal resolutions and can therefore be potentially used for hydrological applications for flash flood forecasting and warning. This study investigates the performance of the rainfall products obtained from weather radar, the Hydro-Estimator (HE) algorithm of NOAA/NESDIS and the Weather Research and Forecasting (WRF) model, and their use in the Hydrologic Engineering Center’s Hydrological Modeling System to simulate the catastrophic flood events that occurred in the Ayamama Basin in Istanbul, Turkey during 7–12 September 2009. The precipitation estimates at 4 km were evaluated against gridded precipitation obtained from 34 rain gauges in the study area. The HE, WRF, and radar precipitation showed capabilities in capturing the timing of the flood events and, to some degree, the spatial distribution. They all, however, underestimated the magnitude of the heavy rainfall which caused the flooding. Among these products, the HE showed the best performance, yielding the lowest bias and root mean squared error. Correlations for radar and HE precipitation were close to each other and substantially greater than that of the WRF model. Traces of flood water available for a portion of the Ayamama basin showed that the 9 September flood event is categorized as an event greater than the 500-year return period, with peak discharge of 253.8 m³/s. Among the flood hydrographs produced using radar, WRF, HE, and rain gauge precipitation inputs, only the gauge-based precipitation produced a flood peak (264.7 m³/s) greater than the 500-year return period on the morning of 9 September 2009. The HE precipitation produced a 100-year flood value (160 m³/s), while radar and WRF precipitation both yielded a 25-year flood value (150 m³/s) on the flood day. Given the poor drainage capacity of the Ayamama basin, however, this 25-year peak discharge value is also capable of causing flooding in the area.