: Drought, is considered as shortage of precipitation comparing to average of it over a long time; that would occur in any kinds of climates. Recently years this phenomenon has changed to a critical problem and has caused great amount of damages. Prediction of drought is the most important solution in contrast with drought. In the last decades, low flow frequency analysis has been considered as one of the methods for prediction of droughts. Shortage of data especially in watersheds without hydrometric stations makes this task difficult to accomplish. Regional analysis is using to reduce these problems. Considering this fact that drought happens regionally and covers large areas and lasts long periods; it is important to perform regional analysis. This study was performed on northern Karoon watershed to predict hydrologic drought. The data of 14 stations suitable for low flow studies with long enough duration was used. Then 13 physiographic and climatic characteristics of chosen stations were selected. Cluster analysis was used to perform homogeneity test for this region. This analysis showed that by elimination of one station (Barez station) this region can be considered as a homogenous watershed. Then seven day low flow time series were calculated for all the remaining stations. Chi-square and Kolmogorov-smirnov best fit tests showed that the two parameter Gamma distribution is the best regional distribution for this region. Therefore seven day low flow was estimated using HYFA for 2, 5, 10, 20, 50 and 100 years. Then low flow regional analysis was performed using 3 methods of multiple regression, low flow index method and distribution parameters regionalization. To compare accuracy of these methods Root Mean Square error (RMSE) statistic was used.
