The design and the project of hydraulic structures, during economic life of the structure, a reliable estimate of the magnitude and the frequency of the maximum flood will occur is of great importance. Regional frequency analysis, being a statistical method, is used for this purpose. Regional flood frequency analysis provides a solution not only in estimating the design event magnitudes but also at sites having too short records and at sites where no hydrologic information is available. In this study, regional flood frequency analysis using L-moments method for estimating the probability distribution parameters were applied to minimum 18 years of recorded series of annual maximum precipitation peaks of 53 precipitation stations which operated by Turkish State of Meteorological Service and Public Waterworks Administration. The index flood method based on L-moments method was used for identification of homogenous region. Generalized Normal, Generalized Extreme Value, Generalized Logistic, Generalized Pareto, Pearson Type 3 and Wakeby distributions were applied to homogenous regions to obtain the reoccurence values. Median, Hosking, Cunnane, Gringorten, Hazen formulas frequently used in literature were used as plotting position formula. Results of the 3 parameters were used for the comparison measurement error. These three parameters are respectively normalized absolute error (NAE), mean absolute error (MAE) and mean square error (RMSE). According to the measurement error value, Generalized Logistic distribution is found to give most accurate results.