Peroxyacetyl nitrate (PAN) is an important chemical species in the troposphere as it aids the long-range transport of NO_x and subsequent formation of O₃ in relatively clean remote regions. Over the past few decades observations from aircraft campaigns and surface sites have been used to better understand the regional distribution of PAN. However, recent measurements made by satellites allow for a global assessment of PAN in the upper troposphere–lower stratosphere (UTLS). In this study, we investigate global PAN distributions from two independent retrieval methodologies, based on measurements from the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) instrument, on board Envisat from the Institute of Meteorology and Climate Research (IMK), Karlsruhe Institute of Technology, and the Department of Physics and Astronomy, University of Leicester (UoL). Retrieving PAN from MIPAS is challenging due to the weak signal in the measurements and contamination from other species. Therefore, we compare the two MIPAS datasets with observations from the Atmospheric Chemistry Experiment Fourier transform spectrometer (ACE-FTS), in situ aircraft data and the 3-D chemical transport model TOMCAT. MIPAS shows peak UTLS PAN concentrations over the biomass burning regions (e.g. ranging from 150 to  >  200 pptv at 150 hPa) and during the summertime Asian monsoon as enhanced convection aids the vertical transport of PAN from the lower atmosphere. At 150 hPa, we find significant differences between the two MIPAS datasets in the tropics, where IMK PAN concentrations are larger by 50–100 pptv. Comparisons between MIPAS and ACE-FTS show better agreement with the UoL MIPAS PAN concentrations at 200 hPa, but with mixed results above this altitude. TOMCAT generally captures the magnitude and structure of climatological aircraft PAN profiles within the observational variability allowing it to be used to investigate the MIPAS PAN differences. TOMCAT–MIPAS comparisons show that the model is both positively (UoL) and negatively (IMK) biased against the satellite products. These results indicate that satellite PAN observations are able to detect realistic spatial variations in PAN in the UTLS, but further work is needed to resolve differences in existing retrievals to allow quantitative use of the products.