Unit hydrograph is one of flood design method that commonly used in Indonesia. Problems emerge in the unit hydrograph application including high rainfall variability both time and space, lack of automatic rainfall data needed for unit hydrograph derivation. However, the output of the catchment system called hydrograph that represent all the phenomena in the system is available. So, effect of rainfall variability on unit hydrograph and the possibility to derive unit hydrograph based on hydrograph data without rainfall data need to be studied. Unit hydrograph derivation is done by using different rainfall input i.e. single station, rainfall catchment and inverse effective rainfall derived based on hydrograph alone. The resulted unit hydrographs are then used for design flood calculation. Results show that the resulted unit hydrographs vary among the different rainfall inputs. The difference in peak relative to unit hydrograph derived based on average rainfall for Bedog, Code and Winongo catchments could reach -18%, -30% and -11%, respectively. This result indicates that in the catchment studied, the rainfall variability is relatively high. However, the difference in flood peak is quite small, that is up to -12% for all the catchments. There is a trend that the station closest to the catchment centroid gives small bias in design flood relative to the design flood computed based on average rainfall.