An advanced runoff model is compared to a simple one employing only a runoff coefficient and a regression parameter allowing for initial loss. The present study shows that the more detailed description of the runoff processes cannot be justified due to the uncertainty from using only one gauge in a catchment for the description of the rain input. A significant variation of the two parameters from one catchment to another has been found and the uncertainty of the two variables are evaluated. The uncertainty of the hydrological reduction factor and the initial loss should be taken into account in the evaluation of the yearly discharges. In the case of extreme events, the uncertainty of the predicted runoff of the single event should also be taken into account.