Taking low-cost sounding rockets as the research object, a multi-disciplinary simulation model for sounding rockets was established on the basis of various disciplinary simulation modules, enabling the simulation of rocket flight performance with multi-disciplinary coupling. An uncertainty propagation analysis method based on dynamic expansion sampling and surrogate models was proposed to address the uncertainty propagation issues related to the flight performance of the sounding rocket. An uncertainty deviation model for rocket flight performance was established on the basis of physical analysis. The deviation parameters were dynamically sampled using the bounded augmented Latin hypercube design method, and deviation samples that satisfied the specified distribution were obtained through the inverse cumulative distribution transformation method. An improved expanded radial basis function model was employed for the approximate modeling of flight performance characteristic parameters, and an approximate prediction model for the flight performance characteristic parameters of sounding rockets was established by utilizing a minimal number of sample points. The flight performance characteristic parameters of rocket obtained from the proposed method were compared with those from the Monte Carlo simulation method. The results validate that the proposed method can achieve rapid and accurate statistical prediction of flight performance parameters by utilizing a minimal number of simulation samples under the specified distribution deviation model.