Abstract: In order to evaluate and reduce the input ground motion uncertainty in the assessment of structural collapse vulnerability, a ground motion uncertainty analysis method based on Lasso regression (Least absolute shrinkage and selection operator, Lasso) is proposed. The ground motion set is composed of ground motion records selected based on the target spectrum matching method, and the ground motion intensity measure (IM) set commonly used in seismic engineering research is selected. The effective IM for predicting the collapse capacity of structures is selected based on the Lasso regression, followed by correlation analysis of the selected IM. The key features are extracted as the clustering basis, and the ground motion clustering is carried out on the original ground motion set. According to the quantitative method of ground motion uncertainty in structural seismic collapse vulnerability analysis based on random sampling, the ground motion clustering results are used as the sampling basis for random sampling to quantify the influence of ground motion uncertainty in structural seismic collapse vulnerability assessment. The results show that compared with the uncertainty quantitative analysis results of the two control groups without using the uncertainty quantitative method in this paper, the uncertainty quantitative values of the structures with different story heights are reduced by more than 50%. The method proposed in this paper can effectively reduce the influence of ground motion uncertainty in vulnerability analysis. The seismic collapse fragility analysis of structures with the same accuracy at the same time can reduce the amount of calculation by half and save computing resources.