Aimed at the problems that a lot of problems remains to be solved by cubature Kalman filter in nonlinear target tracking when the noise covariance is uncertain, an optimized adaptive cubature Kalman filter is proposed. First, the noise covariance is obtained by the linear matrix equation derived from the innovation sequence and the residual sequence, a new process noise covariance [WTHX]Q[WT] estimation method is derived based on the correlation between the innovation sequence and the residual sequence, and then the estimation of the measured noise covariance is made by adopting the residual sequence, and the weighting factors are utilized for combining the current noise covariance matrix and the estimated value into a new measurement noise covariance matrix [WTHX]R[WT], thus avoiding effectively the limitations of inaccurate state estimation. The simulation results show that under conditions of timevarying noise covariance, the tracking accuracy is improving obviously by the proposed adaptive algorithm cubature Kalman filter.