We examine volume mixing ratios (vmr) retrieved from limb emission spectra recorded with the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) on board Envisat. In level 2 (L2) data products of three different retrieval processors, which perform one dimensional (1-D) retrievals, we find significant differences between species' profiles from ascending and descending orbit parts. The relative differences vary systematically with time of the year, latitude, and altitude. In the lower stratosphere their monthly means can reach maxima of 20% for CFC-11, CFC-12, HNO3, H2O, 10% for CH4 and N2O. Relative differences between monthly means of 1-D retrieval results and of the true atmospheric state can be expected to reach half of these percentage values, while relative differences in single vmr profiles might well exceed those numbers. Often there are no physical or chemical reasons for these differences, so they are an indicator for a problem in the data processing. The differences are generally largest at locations where the meridional temperature gradient of the atmosphere is strong. On the contrary, when performing the retrieval with a tomographic two dimensional (2-D) retrieval, L2 products generally do not show these differences. This suggests that inhomogeneities in the temperature field, and possibly in the species' fields, which are accounted for in the 2-D algorithm and not in standard 1-D processors, may cause significant deviations in the results. Inclusion of an externally given adequate temperature gradient in the forward model of a 1-D processor helps to reduce the observed differences. However, only the full tomographic 2-D approach is suitable to resolve the horizontal inhomogeneities. Implications for the use of the 1-D data, e.g. for validation, are discussed. The dependence of the ascending/descending differences on the observation strategy suggests that this problem may affect 1-D retrievals of infrared limb sounders, if the line of sight of the instrument has a significant component in the direction of the horizontal temperature variation.