In order to explore the applicability of the quantile regression model in non-stationary frequency analysis, the quantile regression model and the GAMLSS model were used to analyze the non-stationary frequency of the annual maximum flood series at five stations in the Weihe River Basin and the Pearl River Basin. Qualitative analysis of the probability coverage shows that the goodness of the fitting effect for the quantile regression model is better than that of the GAMLSS model. Quantitative analysis of the Filliben correlation coefficient shows that the GAMLSS model has a better fitting effect but the advantages are not obvious. Comprehensive analysis shows that the overall fitting goodness of the quantile regression model is better than the GAMLSS model.