Abstract: To evaluate the relative influence of precipitation data, model structure and model parameters on runoff simulation uncertainty, this study selected four multiple precipitation datasets (station observations, CRUJRA, CMFD, ERA5-Land gridded data) and constructed a modular component-based hydrological model. The dynamic Sobol' sensitivity analysis method was applied to quantify the sensitivity of input data, model structure, and parameters across 51 sub-basins in the Upper Yangtze River basin. The results are as follows: ①Model parameters contribute most significantly to runoff simulation uncertainty, with the median of multi-year average total order sensitivity index being 0.59 across sub-basins, followed by model structure (0.51) and precipitation data (0.36). ②The interflow process exhibits the highest sensitivity, followed by evapotranspiration, snow balance, surface runoff and baseflow processes. The sensitivities of hydrological sub-processes demonstrate seasonal variability, with the snow balance process being the most pronounced. ③Parameters related to snow balance, evapotranspiration, and interflow processes are critical parameters influencing the runoff simulation uncertainty. ④Mean temperature, elevation, and hydraulic conductivity are significant factors influencing model structure and parameter sensitivities of snow balance, evapotranspiration, and interflow processes. The study identifies time-varying sensitivity patterns of precipitation data, model structure and parameters on runoff simulation uncertainty, providing insights for optimizing hydrological model structure and parameterization.