The risk of natural disasters, such as typhoon, drought, heavy rainfall, flooding and etc. has increased due to abnormal climate and climate changes. These disasters caused huge damages to people and properties. There is the difficulty to predict and prevent the natural disaster due to the lack of observed hydrological data, including runoff and flood peaks. The mid·small catchments, which are a large part of the national land, are ungauged catchments, and it is limited to obtain and manage the observed hydrological data. Research on prediction of hydrological data such as floods and runoff has been carried out for rainfall-runoff analysis at ungauged catchments. However, the current studies are insufficient to apply to the field of water resources. This study developed a regionalised Rainfall-Runoff Model to enhance the accuracy of estimating the runoff at ungauged catchments. This model was developed using runoff characteristics at ungauged catchments in Geum River. The daily hydrological data (rainfall, runoff, evapotranspiration and catchment characteristics) was collected from 2001 to 2015 at 19 catchments in Geum River. Also, in order to evaluate the applicability of the conceptual Rainfall-Runoff Model. Nine models combining MP, CWI, PDM models of the soil moisture accounting and three catchment routing models were applied. Optimized parameters were selected by catchment using the Monte Carlo method for model calibration, and the applicability of parameters was verified by validation. Based on the result of calibration and validation of the nine conceptual rainfall-runoff models, the PDM-2PAR model was selected for the representative optimum model in Geum River catchments. The statistical multiple linear regression equations were developed based on the catchment characteristics and optimized parameters of this model. These equations are regionalised models of parameters in NSE*, RMSE-FH and RMSE-FL objective functions, and were applied to Cheongseong catchment assumed as an ungauged catchment to evaluate the applicability. The results showed that the regionalised models were acceptable with the values of 0.28, 5.79 and 0.13 by NSE*, RMSE-FH and RMSE-FL, respectively. The future studies will include more catchments to improve the statistical regionalised models in predicting runoff in Geum River catchments.