The theoretical characterization of the specific relation between a measurement and the physical quantities we want to determine, along with an assessment of the accuracy of the instrument, form the instrument model. In order to check whether the instrument model is correct, the measurements need to be compared with independent measurements whose characterization and error analysis is better understood. A remote sounding instrument provides measurements that are indirectly related to the atmospheric state. In this paper a general technique for comparing indirect with direct measurements is discussed, applicable to remote sounding measurements obtained using both optimal estimators and maximum likelihood methods. A specific application to validation of Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) ozone retrievals using ozonesondes during winter 2002--2003 is described. No significant bias between the two instruments and a confirmation of the theoretical MIPAS accuracy assessment, after a discussion of specific comparisons, were found.