Abstract: Nonlinear filtering algorithms must be applied to single observer passive location and tracking system for the nonlinearity of its observation equations. MGEKF and other nonlinear filters are belong to EKF in essence, while the EKF-like algorithms share the disadvantage of linearization reduce to the unstability of filters. The filter, based on unscented transformation is named UKF, does not need linearization and shows robustness strongly; however, the convergence of UKF is poor for its underestimation of true covariance. An iterated UKF algorithm is developed, and the estimation to the covariance of filter is improved by iterating estimation. Simulation results demonstrate the tracking performance of this algorithm in different conditions.