We validate the Ozone Monitoring Instrument (OMI) ozone profile (PROFOZ v0.9.3) product including ozone profiles between 0.22 and 261 hPa and stratospheric ozone columns (SOCs) down to 100, 215, and 261 hPa from October 2004 through December 2014 retrieved by the Smithsonian Astrophysical Observatory (SAO) algorithm against the latest Microwave Limb Sound (MLS) v4.2x data. We also evaluate the effects of OMI row anomaly (RA) on the retrieval by dividing the data set into before and after the occurrence of serious RA, i.e., pre-RA (2004–2008) and postRA (2009–2014). During the pre-RA period, OMI ozone profiles agree very well with MLS data. After applying OMI averaging kernels to MLS data, the global mean biases (MBs) are within 3 % between 0.22 and 100 hPa, negative biases are within 3–9 % for lower layers, and the standard deviations (SDs) are 3.5–5 % from 1 to 40 hPa, 6–10 % for upper layers, and 5–20 % for lower layers. OMI shows biases dependent on latitude and solar zenith angle (SZA), but MBs and SDs are mostly within 10 % except for low and high altitudes of high latitudes and SZAs. Compared to the retrievals during the pre-RA period, OMI retrievals during the post-RA period degrade slightly between 5 and 261 hPa with MBs and SDs typically larger by 2–5 %, and degrade much more for pressure less than ∼ 5 hPa, with larger MBs by up to 8 % and SDs by up to 15 %, where the MBs are larger by 10–15 % south of 40◦ N due to the blockage effect of RA and smaller by 15–20 % north of 40◦ N due to the solar contamination effect of RA. The much worse comparisons at high altitudes indicate the UV1 channel of pixels that are not flagged as RA is still affected by the RA. During the pre-RA period, OMI SOCs show very good agreement with MLS data with global mean MBs within 0.6 % and SDs of 1.9 % for SOCs down to 215 and 261 hPa and of 2.30 % for SOC down to 100 hPa. Despite clearly worse ozone profile comparisons during the post-RA period, OMI SOCs only slightly degrade, with SDs larger by 0.4–0.6 % mostly due to looser spatial coincidence criteria as a result of missing data from RA and MBs larger by 0.4–0.7 %. Our retrieval comparisons indicate significant bias trends, especially during the post-RA period. The spatiotemporal variation of our retrieval performance suggests the need to improve OMI’s radiometric calibration to maintain the long-term stability and spatial consistency of the PROFOZ product.