Abstract: Because the selected values of input parameters of generalized Pareto distribution （GPD） model are difficult to avoid uncertainty, the input parameters uncertainty of this model may lead to uncertainty in the seismic hazard estimation. In this paper, we selected northeastern Tibetan Plateau as the case studied area, and proposed an uncertainty analysis process and method of seismic hazard estimation based on the global sensitivity analysis. First, we used the GPD seismicity model to obtain the results of seismic hazard estimation. And then, we selected starting time of earthquake catalog and magnitude threshold to be the input parameters of seismicity model. The E-FAST method with global sensitivity analysis function was used to quantitatively analyze the influence of the uncertainties of the two parameters and the interaction between the two parameters on the uncertainty of seismic hazard estimation. The results show that the seismic hazard estimation of the GPD model is more sensitive to the magnitude threshold. With different return periods, the sensitivity degree of seismic hazard estimation to magnitude threshold is different. For different return periods, there are nonlinear effects between the two input parameters on the uncertainty of seismic hazard estimation, and the degree of nonlinear effects is different. The uncertainty analysis process and method proposed in this paper can be applied to the uncertainty analysis of seismic hazard estimation based on other seismicity models.