Retrieval of abundances of atmospheric species from limb infrared emission spectra requires accurate knowledge of the pointing of the instrument in terms of elevation, as well as temperature and pressure profiles. An optimal estimation-based method is presented to infer these quantities from measured spectra. The successful and efficient joint retrieval of these largely correlated quantities depends strongly on the proper selection of the retrieval space, the selection of spectral microwindows, and the choice of reasonable constraints which force the solution to be stable. The proposed strategy was applied to limb emission spectra recorded with the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) on board the Envisat research satellite in order to validate the instrument pointing information based on the satellite's orbit and attitude control system which uses star tracker information as a reference. Both systematic and periodic pointing calibration errors were detected, which meanwhile have been corrected to a major part. Furthermore, occasional pitch jumps were detected, which could be assigned to parameter uploads to the satellite's orbit and attitude control system. It has been shown that in general, it is justified to assume local thermodynamic equilibrium below 60 km for these purposes. The retrieval method presented has been proven to be suitable for independent monitoring of MIPAS line-of-sight pointing.