This study investigated the relationship between the errors involved in the areal average rainfall estimation and the runoff simulation (hereafter, rainfall and runoff errors). The error statistics of the observed areal average rainfall were estimated and used for the generation of input data for the runoff simulation. The Clark instantaneous unit hydrograph was used for this runoff simulation. The runoff model parameters were estimated using several sets of rainfall-runoff data observed, whose statistics were then used for the sensitive analysis of the model simulation on the parameter sets. The rainfall error in this study was defined as the relative difference between the observed and generated areal average rainfall. On the other hand, the runoff error was designed to consider the runoff volume, peak flow and peak time, respectively. This study was applied to the Chugnju Dam Basin, Korea. The results obtained are as follows: (1) The variation of model parameters estimated in this study were about 30% of their means. Also, the effect of this error in model parameter estimation on runoff simulation was found to be maximum 15% of the peak flow. (2) The estimation error of areal average rainfall was found roughly proportional to the areal average rainfall itself. Both the rainfall and runoff errors were found to have no obvious biases. However, the variance of the peak flow error was found to be significantly higher. (3) The relationship between rainfall error and runoff volume error was roughly one to one, however, the rainfall error has become amplified by more than 50 % and transferred to the peak flow error.