Abstract: To apply the method of high-order L-moments (LH moments) to parameter analysis of precipitation frequency, this paper formulates LH moments of Pearson type III distribution through mathematical processing. Monte-Carlo simulations were conducted to examine the statistical properties of the L-moment methods of different orders and their influences on the design values of different return periods. We estimated parameters of the P-III distribution and LH moments for six annual precipitation series of Shaanxi province and evaluated the effect of fitting large sample events and its calculation error. The results show that the LH moments describe the data series better than the commonly used L-moments. An appropriate order of L-moment is useful for the design values of long return period.