The long- and short-term runoff model is one of the valuable runoff analysis in moutainous basins. This model contains advantages in each of the long-term and flood runoff models. In other words, it has characteristics that calucurate effective rainfall inside it and analyze both flood and long-term runoff successively. In this paper, the long- ans short-term runoff model was applied to runoff analysis in the Ochozu study basin of 0.095 km2 and applicability of it to the small mountainous basin. Characteristics between it and TOPMODEL were discussed through the prediction accuracy of river discharge. Tourteen unknown parameters in the model were identified using a simple genetic algorithm (SGA) with the fitness function defined by a x2 error between observed and caluclated river discharges. The results indicated that the SGA-optimized the long- and short-term runoff model and could successfully predict the river discharge in rainy season.