GOMOS ozone profiles were analyzed in a joint contribution of the Ground-based Measurement and Campaign Database (GBMCD) subgroup. In the analysis study 131 collocated ozone profiles of ground-based lidar systems, microwave radiometers, and balloon sondes were used for the validation. We have distinguished between three different parameters which might influence the GOMOS data quality. The pairs of collocated profiles were separated by (1) brightness of the limb during the GOMOS observation, and (2) the magnitude value and (3) temperature of the observed star. For each selection the mean difference between the GOMOS and GBMCD ozone profile was calculated. The GOMOS retrieved ozone profile is strongly affected by the brightness of the limb in which the star occults. Bright limb situations give poor results. Although, in this situation there is an exception for stars with a magnitude value smaller than 1. In that case the results are reasonable between 30 and 50-km altitude, but GOMOS is lower by 10-15%. Twilight limb conditions give better results, but there are still large deviations and it needs further research. Good results come from ozone profiles measured in dark limb situation. Then the bias between 18 and 45-km altitude is within 5 to 10%. The ozone profiles between 45 and 65-km altitude, measured in the dark limb using cold stars, give poor results, but using onldy hot stars results in a bias lying within 20%. In this case though, it is a significant non-random bias and this suggests a possibility for improvement.