Aiming at the uncertainties of various conditions and tasks in the reentry gliding vehicle, an local model based on-line trajectory adjustment algorithm is proposed by the combining of off-line trajectory optimization and on-line trajectory adjustment. Off-line Radau-pseudospectral solution is used for trajectory characterize and local trajectory model construct, with the local model set constructed by the trajectory constraint space discretization. A method of local model subset construct and updating based on fuzzy membership degree is proposed by introducing the idea of fuzzy clustering, with the on-line generation of trajectory realized by the means of trajectory weighted fusion. Finally, the trajectory generation and on-line adjustment performance of the algorithm are verified by simulation analysis.