This is a presentation of the techniques used to process inertial and GPS data collected during an airborne gravimetric and altimetric survey in order to obtain absolute position and attitude estimates of the sensors mounted on board the aircraft. For high precision gravimetry and altimetry, these navigational parameters are required, in the order of a decimeter for position and better than one arc-minute for attitude. The survey took place in the Portuguese Azorean area in October 1997. It consisted of flying a series of profiles with a set of instruments mounted in a military aircraft, covering a large area around the islands, mainly over the mid-Atlantic ridge. The instruments whose position and attitude had to be determined were a gravimeter and two altimeters (one laser and another radar). For this purpose, three GPS receivers and an Inertial Measurement Unit (IMU) were used. The basic idea was to have a relatively low cost system that would be able to supply position and attitude with a level of accuracy and reliability similar to other more expensive systems, like gimbaled and ring laser gyro based inertial systems. One other requirement was that it should be able to work in real time. This feature could be useful for repeating flight tracks and for many other applications. For this purpose, a radio link was established with the aircraft to supply GPS data from a ground reference station. The methodology for processing GPS and IMU raw data was implemented with an integrated feedback loop. The position determination is based on real time DGPS with L1 and L2 carrier phase ambiguity fixing, using online IMU processed data to validate and increase the positions reliability, especially during GPS outages and when the aircraft is far away from the nearest GPS reference station. For the attitude angles, the IMU raw data is processed using the absolute GPS position measurements for continuous alignment. The baseline between a pair of GPS antennas is also computed (without resorting to a reference station data), using L1 and L2 carrier phase and doppler measurements. This baseline is very helpful for the continuous alignment of the strapdown analytical platform in heading, since this angle estimate is weakly coupled to position updates, especially in the absence of high horizontal accelerations. The position and attitude values were obtained from all the measurements using an Extended Kalman filter. Besides the normal navigation states, the filter also includes extra states to account for the inertial sensor errors and the vertical gravity anomaly.