Kalman filter is not an optimal estimation method for the systems without any exact model. Therefore, multiple fading factors matrix has been used as a multiplier for the covariance matrices in these systems. In this paper, a novel method named adaptive multiple fading factor Kalman filter is proposed for the systems without initial alignment of strap down inertial navigation systems (SINS). By applying this algorithm to different channels of Kalman filter, different coefficients of fading factors are computed. The simulation results show a noticeable increment in alignment's precision, and alignment's speed, and a noticeable decrement in sensitivity to unknown noises.