Solar occultation measurement of atmospheric species with short photochemical lifetimes poses some difficulties if sharp gradients are present in the species concentrations near sunrise/sunset conditions. These photochemically induced variations introduce asymmetries in the species distribution along the line of sight that need to be taken into account in the retrieval process in order to prevent inaccuracies. Correction factors derived from photochemical model calculations have been routinely applied to the retrievals of stratospheric NO and NO2 by HALOE and ATMOS experiments. Mesospheric ozone, because of its twilight variations, also belongs to the group of species which require a correction procedure. Results from our mesospheric diurnal photochemical model indicate that proper accounting of diurnal variations leads to more than 20% increase in the ozone column along the line of sight for sunrise conditions near 0.1 hPA at the equator in January. Correspondingly, the retrieved ozone at the tangent point will be lower if these twilight variations are considered. The effects under sunset conditions are smaller. The current HALOE retrieval algorithm uses an approximate correction for mesospheric ozone. We have developed a new database of twilight gradients for the entire altitude range of HALOE measurements based on the results from a contemporary diurnal photochemical model. HALOE scans for January 29, 1992, have been reprocessed using the new diurnal correction factors. Sunrise ozone mixing ratios near 0.1 hPa are smaller by more than 20% compared to the HALOE V19 data. The differences between the corrections for sunrise and sunset retrievals suggest that, with the new diurnal corrections, the sunrise to sunset ozone ratios near 0.1 hPa obtained from approximately coincident HALOE data will be in much better agreement with the values derived from photochemical theory.