AbstractTriples of GPS radio occultation (RO) temperature data are used to derive horizontal and vertical gravity wave (GW) parameters in the stratosphere between 20 km and 40 km from which the vertical flux of horizontal momentum is determined. Compared to previous studies using RO data, better limiting values for the sampling distance (Δd≤250 km) and the time interval (Δt≤15 min) are used. For several latitude bands the mean momentum fluxes (MFs) derived in this study are considerably larger than MF from other satellite missions based on horizontal wavelengths calculated between two adjacent temperature profiles along the satellite track. Error sources for the estimation of MF from RO data and the geometrical setup for the applied method are investigated. Another crucial issue discussed in this paper is the influence of different background separation methods to the final MF. For GW analysis a measured temperature profile is divided into a fluctuation and a background and it is assumed that the fluctuation is caused by GWs only. For the background separation, i.e., the detrending of large‐scale processes from the measured temperature profile, several methods exist. In this study we compare different detrending approaches and for the first time an attempt is made to detrend RO data with ERA‐Interim data from the European Centre for Medium‐Range Weather Forecasts. We demonstrate that the horizontal detrending based on RO data and ERA‐Interim gives more consistent results compared with a vertical detrending.