The total amount and time distribution of rainfall are crucial as they significantly influence the peak flow in rainfall-runoff analysis. Mostly, HUFF’s method is being used for rainfall’s time distribution; however, Huff’s method is a time distribution method used for rainfall designed for observed rainfall data for a single rain of above certain threshold, so it utilizes only one summation curve regardless of rainfall’s duration time. Therefore, it cannot properly reflect the characteristics of rainfall distribution according to rainfall’s duration time. In addition, it exhibits a problem that the peak flood is underestimated in the small basin like urban area. This study is intended to solve the biggest problems of Huff’s method, not properly considering the characteristics of time distribution variations for each duration time and applying the criterial rainfall amount for rainfall’s total amount in each of the rainfall events in a lump. Accordingly, we used AWS’ minute-based observational data to build up POT series with annual peak data and developed rainfall’s time distribution model appropriate for the urban basin based on the actual rainfall events that had happened in the past. This study aims to propose a watershed-based rainfall time distribution model applicable not only for the flood forecasting and warning, but also for the hydro-infrastructure design.