Regional flood frequency analysis is one of the most important methods for estimating flood in poorly gauged or ungauged catchments. In this study, a regional flood frequency analysis has been carried out, using cluster analysis approach in Great Karun watershed. After considering the stations in the area, 44 stations which had good data were selected. The watershed was firstly divided into 3 regions based on the physiographic characteristics and then by using the Hosking & Wallis homogeneity criterion and removing or displacing the stations, the watershed was finally divided into 4 homogeneous regions. In order to find the most suitable regional distributions for homogeneous regions, L-moment ratio diagram and Hosking & Wallis goodness of fit test has been used. The distribution parameters were estimated using the method of probability weighted moments (PWMs). Therefore, by calculating the average discharge, the maximum flood discharge for a given return period can be estimated for ungauged area. By applying this method, the regional flood frequency estimation will be possible for every zone, even ungauged area of the great Karun watershed.