In this study, it was aimed to determine regional frequency analysis for selected 26 stations out of 41 streamflow gauging stations located in Sefidrud Dam watershed in north of Iran. 7-day low flows are considered as indices for hydrological drought analysis. The study area was identified as heterogeneous based on the homogeneity measurement (H1=13.68) by a software with code written in FORTRAN language. Therefore Fuzzy Cluster Analysis (FCA) algorithm is applied to distinguish the homogeneous sub regions by taking into consideration the physiographic and hydrologic characteristics of the watershed. Two regions were identified as the result of FCA analysis namely; east region and west region. According to the L-Moment analysis, East region of the watershed was found homogeneous and the West region of the watershed became homogeneous after removing the Gilvan, Yengikend and Firuzabad streamflow stations. The Generalized Logistic (GLO) and Pearson Type III (PIII) distributions were selected as the best fit regional distributions for the east and west sub-regions, respectively. The best fit distribution is selected with the smallest absolute value of the Z statistic (ZDIST) based on the goodness-of-fit-test. The selected best fit distributions are advised to use in the regional low-flow forecasting of Sefidrud watershed.