Because of the existence of significant variation of the environment and vehicle characteristics during the flight of missiles, the flight dynamic characteristics and the structure modal parameters of missiles also vary in a wide range, thus making it hard to model the system accurately. This brings potential danger to the aeroservoelastic stability of missiles. Conventional structure filters are often unsuitable for use. This article presents a new design method of adaptive structure filters for missiles. Sample signals of input command and sensor output are utilized to establish the autoregressive moving average (ARMA) model of missiles. By applying the recursive least square method, realtime structure modal parameters of missiles are identified and utilized to update the central frequency of structure filter. Simulation results indicate that, compared with the conventional structure filter, the adaptive structure filter can increase the stability margin of missiles significantly in the case of wide-range parameter perturbations, and that it is robust with regard to changes in flight environment and structure characteristics.