Abstract: A strong tracking cubature Kalman filter(STCKF) algorithm with disturbing fitting function is proposed for Leo satellite orbit determination when inaccurate initial value and orbit model lead to low precision of filter. The suboptimal fading factor is calculated using the equivalent expression of strong tracking filter(STF) to adjust the gain matrix online and to force the residual sequence orthogonal to each other, which effectively reduces the sensitivity to the initial state. Improved Eular method is used to disperse the orbital dynamic equation with J₂ perturbation, and the polynomial fitting function is used to represent the rest disturbing force. The simulation results show that STCKF with disturbing fitting function can effectively improve the orbit determination accuracy, and reduce the dependence on initial value of the filter.