Recent studies have shown that weak Global Navigation Satellite System (GNSS) signals could potentially be used to navigate from the Earth to the Moon. This would increase autonomy, robustness and flexibility of the navigation architectures for future lunar missions. However, the utilization of GNSS signals at very high altitudes close to the Moon can be significantly limited by the very low power levels seen at the receiver’s antenna. This can result in a strongly reduced visibility of the GNSS satellites, which can worsen the already poor relative geometry of the GNSS receiver to the GNSS satellites. Furthermore, during most of a Moon Transfer Orbit (MTO), the very weak GNSS signals are also affected by Doppler shifts and Doppler rates larger than the ones generally experienced on the Earth, due to the much higher relative dynamics between the receiver and the transmitters. As a consequence, commercial GNSS receivers for terrestrial use cannot successfully acquire and track such signals. More advanced architectures and specific implementations are thus required to use GNSS for lunar missions. In this paper we propose the use of an adaptive orbital filter to aid the GNSS acquisition and tracking modules and to strongly increase the achievable navigation accuracy. The paper describes the orbital filter architecture and tests results carried out by processing realistic radio frequency (RF) signals generated by our Spirent GSS 8000 full constellation simulator for a highly elliptical MTO.