To　improve　the　anti-interference　ability　of　initial　alignment,　mean　algorithm　and　Kalman　filter　algorithm　for　initial　alignment　are　compared,　and　a　data-reuse　method　for　combining　these　two　algorithms　is　proposed.　One　group　of　sampling　data　is　reused　for　coarse　and　fine　alignment,　prolonging　the　coarse　alignment　time　and　making　off-line　algorithms　aviable　for　fine　alignment.　The　method　was　applied　in　a　strapdown　system,　and　experimental　results　show　that　the　method　can　effectively　restain　the　passanger　disturbance.