This study was conducted to develop an optimal runoff bydrograph model by comparison of the peak discharge and time to peak between observed and simulated flows derived by four different models, that is, linear time-invariant, linear time-variant, nonlinear time-invariant and nonlinear time-variant models under the conditions of heavy rainfalls with regionally uniform rainfall intensity in short durations at nine small watersheds. The results obtained through this study can be summarized as follows. 1. Parameters for four models including linear time-invariant, linear time-variant, nonlinear time-invariant and nonlinear time-variant models were calibrated using a trial and error method with rainfall and runoff data for the applied watersheds. Regression analysis among parameters, rainfall and watershed characteristics were established for both linear time-invariant and nonlinear time-invariant models. 2. Correlation coefficients of the simulated peak discharge of calibrated runoff hydrographs by using four models were shown to be a high significant to the peak of observed runoff graphs. Especially, it can be concluded that the simulated peak discharge of a linear time-variant model is approaching more closely to the observed runoff hydrograph in comparison with those of three models in the applied watersheds. 3. Correlation coefficients of the simulated time to peak of calibrated runoff hydrographs by using a linear time-variant model were shown to be a high significant to the time to peak of observed runoff hydrographs than those of the other models. 4. The peak discharge and time to peak of simulated runoff hydrogaphs by using linear time-variant model are verified to be approached more closely to those of observed runoff hydrographs than those of three models in the applied watersheds. 5. It can be generally concluded that the shape of simulated hydrograph based on a linear time-variant model is getting closer to the observed runoff hydrograph than those of three models in the applied watersheds. 6. Simulated hydrographs using the nonlinear time-variant model which is based on more closely to the theoritical background of the natural runoff process are not closer to the observed runoff hydrographs in comparison with those of three models in the applied watersheds. Consequently, it is to be desired that futher study for the nonlinear time-variant model should be continued with verification using rainfall-runoff data of the other watersheds in addition to the review of analyical techniques.