The performance of two nonlinear estimators is compared for the localization of a spacecraft. It is assumed that range measurements are not available (as in deep space missions), and the localization problem is tackled on the basis of angles-only measurements. A dynamic model of the spacecraft accounting for several perturbing effects, such as Earth and Moon gravitational field asymmetry and errors associated with the Moon ephemerides, is employed. The measurement process is based on elevation and azimuth of Moon and Earth with respect to the spacecraft reference system. Position and velocity of the spacecraft are estimated using both the extended Kalman filter (EKF) and the unscented Kalman filter (UKF). The behavior of the filters is compared on two sample missions: Earth-to-Moon transfer and geostationary orbit raising.