Statistical analysis of extreme events is often carried out to obtain the probability distribution of floods data and then predict the occurrence of floods for a significant return period. L-moments approach is known as the most popular approach in frequency analysis. This paper discusses comparison of the L-moments method with higher order moments (LH-moments) method. LH-moment, a generalization of Lmoment, which is proposed based on the linear combinations of higher-order statistics has been used to characterize the upper part of distributions and larger events in flood data. It is observed from a comparative study that the results of the analysis of observed data and the diagram based on the K3D-II distribution using LH-moments method is more efficient and reasonable than the L-moments method for estimating data of the upper part of the distribution events.