Drought is a recurrent extreme climate event with tremendous hazard for every specter of natural environment and human lives. Drought analysis usually involves characterizing drought severity, duration and intensity. Usually, long-term datasets of hydrometric and hydrochemical information are needed to begin an evaluation of dominant low flow (as hydrologic drought indices) producing processes however, in many catchments, these data are not available. A major research challenge in ungauged basins is to quickly assess the dominant hydrological processes of watersheds. In this paper, for developing regional models, low flow analysis has been performed by 3 regression methods (multivariate regression, low flow index method, regionalization model of frequency formula parameters) and Hybrid low flow model in Karkheh basin (southwestern of Iran). Estimated error for four methods show although hybrid method can also use for low flow regionalization analysis but multivariate regression and low flow index methods are more suitable for this purpose.