The approach of ��H∞ filtering is presented, which is applied to alignment of inertial navigation systems. The error model of INS(Inertial Navigation System) is introduced,and the observability of the INS error model is analyzed. ��H∞ filter is designed to an observable subsystem. The simulation results of alignment are given in the condition of bounded noise variance and unknown noise statistics.The results show that ��H∞ filtering is robustness and it is less sensitive to uncertainty in the noise signals statistics than Kalman filtering. It is demonstrated that ��H∞ filtering is efficient to alignment of inertial navigation systems. In order to improve the velocity of alignment, a fast estimation method of the azimuth error angle is proposed for the initial alignment of INS . The simulation results illustrate the efficiency of the method.