Low flow is a natural phenomenon occurring seasonally and is an integral component of the flow regime. The knowledge about low flow values is essential in water management. The best methods of low flow estimation are based on direct – statistical approach. But this kind of evaluation is possible only in case of sites in which there are continuous flow measurements. The problem is when the information is not available or are insufficient in terms of quality and quantity. Recently the most popular method is that based on regionalization, in which the assumption is taken that catchments with similar climate, geology, topography, vegetation and soils would normally have similar streamflow responses for example, in terms of unit runoff, etc. In hydrology approaches which have been used for regionalization of watersheds include: methods of residuals pattern approach, the multivariate statistics such as cluster analysis, Classification and Regression Tree models or seasonality of low flows. In Europe, the first statistical procedures for low flows were developed for catchments in the United Kingdom. Also, the national procedure for low flow estimation depending on available data was made in Austria. In Poland there is no generally established procedure for this issue. The paper focuses on review of methods which are uses in case of hydrological regionalization and are commonly use of low flow estimation in ungauged sites and can be adopted in Polish condition. The review is largely based on the research reporting during last 20 years.