Accurately assessing the robustness of the aerothermal performance of the blade tip is important considering that uncertainty is inevitable in the actual operation of turbines. However, the conventional uncertainty quantification methods are computationally inefficient for such an expensive black-box problem as turbine aerothermal performance prediction. In this paper, an efficient framework that is based on the combination of the sparse polynomial chaos expansion (PCE) and universal Kriging (UK) metamodel is applied to the uncertainty quantification of the effect of the conventional squealer tip and three different winglet squealer tips on the aerodynamic performance of the GE-E3 rotor blade tip. However, the inlet total pressure, inlet total temperature, and inlet flow angle are considered to flow condition uncertainty parameters and tip clearance is considered a geometrical uncertainty parameter. According to the results of the uncertainty quantification, in actual operation, although the setup of the winglet structure can still reduce the leakage flowrate, its effect will be much lower than predicted by deterministic calculations. The parameter that has the greatest influence on the uncertainty of the aerodynamic performance of the four tip structures is the tip clearance. Therefore, the geometric accuracy of the tip clearance should be strictly ensured in the turbine blade assembly and marching process. The uncertainty quantification calculations reveal that there is an antagonistic relationship between the pressure side cavity and suction side cavity on the aerodynamic performance uncertainty of the blade tip, which indicates a reasonable ratio of pressure side cavity and suction side cavity can make the fluctuation of the aerodynamic performance of the pressure side cavity vortex and suction side cavity vortex completely cancel, and thus design the winglet squealer tip with strong aerodynamic performance robustness.