**Abstract** : Ionosphere is a dispersive medium that can strongly affect GPS and GALILEO signals. It is the largest source of ranging error in GNSS. In future GNSS civil aviation context, to remove this effect from pseudoranges, it is necessary to use two different frequencies to obtain "ionospheric-free" measurements, in a dual frequency mode of operation. A receiver can lose one or more frequencies, for instance in the case of disturbance due to RFI leading to the use of only one frequency to estimate ionospheric delay in a degraded mode of operation. Therefore, it is felt by the authors as an important task to identify and determine the performance of techniques that would try to sustain multi-frequency ionospheric delay estimation performance when a multi-constellation receiver installed in an aircraft is losing one frequency component, during critical phases of flight. This problem is identified for instance in [NATS, 2003]. Those single frequency techniques can be based on the fact that the ionospheric delay encountered in GNSS may be estimated by using the remaining code and phase measurements thanks to the dispersive effect of the ionosphere on GNSS electromagnetic waves crossing this medium. 1148 ION GNSS 21st. International Technical Meeting of the Satellite Division, 16-19, September 2008, Savannah, GA. Indeed, ionosphere generates a delay in code measurements as well as an advance in carrier phase with the same amplitude. Making the difference between the two measurements at the same frequency, it is consequently possible to estimate ionospheric delay instead of using "ionospheric-free" pseudorange measurements as in a dual frequency mode. Nevertheless, carrier phase ambiguities must be estimated before estimating single frequency ionospheric delay. Indeed, the difference between code and carrier phase measurements provides twice the ionospheric delay plus phase ambiguity, residual noise and multipath. However, carrier phase measurements are subject to cycle slips that would result in a variation of the phase ambiguity and so in an additive error on the ionospheric delay estimation. Consequently, cycle slips in carrier phase measurements must be monitored to comply with civil aviation requirements to ensure integrity of the system. This is done in [Ouzeau, 2006], the availability of such a technique is also discussed and evaluated in [Ouzeau, 2007]. To estimate code minus carrier ionospheric delay, a Kalman filter is implemented, whose states are both ionospheric delay and carrier phase ambiguities. This allows estimating ionospheric delay and monitoring cycle slips thanks to ambiguities for each satellite in view [Ouzeau, 2007]. During simulations on actual measurements, the filter behavior is studied and the accuracy of the estimation is discussed. This algorithm is expected to bridge a gap between one nominal mode of operation and a degraded mode and thus to try to maintain the level of performance during the degraded mode as long as possible after the degradation occured. In particular, in this paper, the accuracy of the technique is studied. The particular case of APV phase of flight requirements is discussed and actual aircraft measurements are used to validate the model and to observe the filter behavior under actual conditions. The main goals of this paper are then to describe the methodology used to estimate ionospheric delay and in particular the settings of the Kalman filter used, and to present the accuracy of the ionospheric delay estimation obtained. The paper starts with a description of the proposed algorithm and the settings of the Kalman filter. The accuracy of single and dual frequency estimations will be then compared. Ionosphere is a dispersive medium that can strongly affect GPS and GALILEO signals. It is the largest source of ranging error in GNSS. In future GNSS civil aviation context, to remove this effect from pseudoranges, it is necessary to use two different frequencies to obtain "ionospheric-free" measurements, in a dual frequency mode of operation. A receiver can lose one or more frequencies, for instance in the case of disturbance due to RFI leading to the use of only one frequency to estimate ionospheric delay in a degraded mode of operation. Therefore, it is felt by the authors as an important task to identify and determine the performance of techniques that would try to sustain multi-frequency ionospheric delay estimation performance when a multi-constellation receiver installed in an aircraft is losing one frequency component, during critical phases of flight. This problem is identified for instance in [NATS, 2003]. Those single frequency techniques can be based on the fact that the ionospheric delay encountered in GNSS may be estimated by using the remaining code and phase measurements thanks to the dispersive effect of the ionosphere on GNSS electromagnetic waves crossing this medium. 1148 ION GNSS 21st. International Technical Meeting of the Satellite Division, 16-19, September 2008, Savannah, GA. Indeed, ionosphere generates a delay in code measurements as well as an advance in carrier phase with the same amplitude. Making the difference between the two measurements at the same frequency, it is consequently possible to estimate ionospheric delay instead of using "ionospheric-free" pseudorange measurements as in a dual frequency mode. Nevertheless, carrier phase ambiguities must be estimated before estimating single frequency ionospheric delay. Indeed, the difference between code and carrier phase measurements provides twice the ionospheric delay plus phase ambiguity, residual noise and multipath. However, carrier phase measurements are subject to cycle slips that would result in a variation of the phase ambiguity and so in an additive error on the ionospheric delay estimation. Consequently, cycle slips in carrier phase measurements must be monitored to comply with civil aviation requirements to ensure integrity of the system. This is done in [Ouzeau, 2006], the availability of such a technique is also discussed and evaluated in [Ouzeau, 2007]. To estimate code minus carrier ionospheric delay, a Kalman filter is implemented, whose states are both ionospheric delay and carrier phase ambiguities. This allows estimating ionospheric delay and monitoring cycle slips thanks to ambiguities for each satellite in view [Ouzeau, 2007]. During simulations on actual measurements, the filter behavior is studied and the accuracy of the estimation is discussed. This algorithm is expected to bridge a gap between one nominal mode of operation and a degraded mode and thus to try to maintain the level of performance during the degraded mode as long as possible after the degradation occured. In particular, in this paper, the accuracy of the technique is studied. The particular case of APV phase of flight requirements is discussed and actual aircraft measurements are used to validate the model and to observe the filter behavior under actual conditions. The main goals of this paper are then to describe the methodology used to estimate ionospheric delay and in particular the settings of the Kalman filter used, and to present the accuracy of the ionospheric delay estimation obtained. The paper starts with a description of the proposed algorithm and the settings of the Kalman filter. The accuracy of single and dual frequency estimations will be then compared.