Flood disaster generates large amounts of flood wastes and it is rarely well-treated due to the heavy amount of wastes.Therefore the prediction of flood waste generation and making a plan for appropriate treatment have been required.Targeting for Seoul city, we investigated the history of flood disasters and the number of flooded buildings, and conductedregression analysis between the number of flooded buildings and rainfall data. As a result, daily average rainfall andmaximum daily rainfall showed better correlations with the number of flooded buildings than total rainfall data. As aresult of verification on the official unit of flood waste generation, 1.35ton/flooded building, it was reliable data for totalSeoul city, but not for regional units (gu). Finally, using regression equation and the official unit of flood waste generation,we suggested the method to predict flood waste generation by rainfall. For more reliable prediction, it is necessary toaccumulate more regional data about the number of flooded buildings, building types, and topographic conditions dueto a wide regional deviation.