With an application to target tracking with unknown process noise, adaptive UKF is presented. In this new algorithm, modified Sage-Husa noise statistics estimator is introduced to estimate the system process noise variance adaptively. By estimating the noise covariance online, the proposed method is able to compensate the errors resulting from the change of the noise statistics. Such a mechanism can improve the state estimation accuracy and enlarges its application scope. The simulations show that adaptive UKF can provide better performance in tracking accuracy than the standard UKF, especially in the case of unknown prior system noise statistics.