Discontinuous systems have been increasingly paid attention since they can be found ranging from physical to biological systems. In this paper we consider an effective estimation algorithm for systems with discontinuous vector field. We propose an incorporation between the existing recedinghorizon nonlinear Kalman filter (RNKF) and the unscented transformation, which is named receding-horizon unscented Kalman filter (RUKF). The use of unscented transformation is beneficial to discontinuous systems since it does not require partial derivatives as does the linearization technique which may incur severeness at discontinuity. An application of this algorithm to a system with discontinuous friction is considered to illustrate its performance in comparison with the classical unscented Kalman filter (UKF).