Due to global climate change and the growing impact of human activity,frequency and intensity of extreme precipitation events in the Pearl River basin are changed,and the consistency of non-stationary extreme precipitation no longer be guaranteed under changing environment. This paper attempts to explore the spatiotemporal variations and the causes of non-stationary extreme precipitation using time-varying Generalized Pareto distribution (GPD) based on daily rainfall of 43 sites during 1960-2012 in the Pearl River basin. The results show that the extreme value index of extreme precipitation sequence decreases gradually from east to west, which indicates that the correlations of the extreme precipitations reduce from east to west. The region with large variationof extreme precipitation at present tends to weaken,and on the contrary,the region with small variation of extreme precipitation at present tends to strengthen. The amount of 50- and 100-years return period daily precipitation in the Pearl River delta,the southern of the Dongjiang River basin and the northeastern of Liujiang River basin is greater than that in the western region of the Nanpan River basin. The SME index of ENSO is the most significant factor of extreme precipitation among the 7 factors. The comparative result of the time varying POT model and stationary POT model shows that the time varying POT model of non-stationary proposed in this paper can better overcome the problem of non-stationary characteristics in parts of the study basin.