Ward's hierarchical clustering method is applied to classify the annual floods and identify the hydrologic homogeneous regions in Turkey. For this aim, the annual flood data obtained at the 117 gauging stations having data records of 31 years operated by the general directorate of electrical power resources survey and development administration (EIE) throughout Turkey are considered. Discordancy and regional homogeneity measures are applied to test homogeneity of regions identified by Ward's cluster method. Flood frequency analyses for seven sub-groups defined by Ward's clustering method are carried out using various frequency distributions based on index flood and L-moments approaches. The best fit distributions for all sub-regions are identified based on L-moments goodness of fit statistic. The accuracy of results of quantile estimates are evaluated by using relative RMSE% and relative BIAS% through the use of Monte Carlo simulation. (C) 2014 Sharif University of Technology. All rights reserved.