In initial alignment of strap-down inertial navigation system(SINS),because of the parameters model have some deviations from the real system,and the noise statistic characteristics is not exactly known,the regular Kalman filter can not gain the ideal consequence.For avoiding filter divergence and model incertitude,an interval adaptive Kalman filter(IAKF) based on Sage-Husa algorithm is proposed,exactitude error model of SINS and the interval adaptive Kalman filter equation are given.When the noise statistical characteristics are unknown,comparing the usage between regular Kalman filter and interval adaptive Kalman filter in initial alignment.The simulation result show that the algorithm can improve system effectiveness,it is an ideal navigation filter algorithm of initial alignment.