In this study a new method has been proposed which was developed for detecting outliers and calculating flood discharge. In the study the data of six gauging stations on Euphrates basin have been used. The data of these stations were analyzed and the outliers were detected by using Rosner test, Dixon Thompson test and the proposed method. The flood discharge which can occur in a certain recurrence interval has been calculated by using the proposed method. After the outliers were removed from the datasets, the flood discharges were calculated by using probability density functions (PDF) which were determined by using goodness of fit tests. In addition, flood discharges are calculated regionally by using PDF whose parameters were determined by the L moments method. When the flood discharge values were calculated by using the proposed method, and when the flood frequency analysis and L moments methods were compared, it has been concluded that the values determined by the proposed method and flood frequency analysis are more suitable and usable than the values determined by the L moments method.