Since the discovery of the stratospheric ozone depletion in the 1980s [World Meteorological Organization/United Nations Environment Programme (WMO/UNEP), 1994; Solomon, 1999; Staehelin et al.; 2001], the interest of its measurement has substantially increased. Indeed, this gas highly absorbs the dangerous UV radiation and preserves the biosphere. Long series of measurements of total ozone column and of solar UV radiation are needed to define climatologies of both UV and ozone and to study the impact of the ozone depletion at the Earth surface. Total ozone column is determined at Villeneuve d'Ascq, France, by a differential absorption method, using global spectral UV measurements performed with a Jobin Yvon spectroradiometer by the Laboratoire d'Optique Atmospherique. This method is based on the comparison between two irradiance ratios, one simulated beforehand and stored in a lookup table (LUT) and one calculated from the measured spectrum [Stamnes et al.; 1991]. This paper presents the method used to improve the choice of the different irradiances used to calculate the ratios and presents the five irradiance ratios that are finally kept. Some sensitivity tests are performed in order to find the validity range of a LUT and the number of LUTs needed. These tests reveal that the aerosol Ångström coefficient, the aerosol optical depth, and the atmospheric vertical profiles may induce errors in the ozone retrieval. That makes it necessary to calculate 21 LUTs and to select one of them depending on the season and on the aerosol properties. Then an estimation of the uncertainty of the method is performed, enabling us to conclude that the method leads to approximately 3% relative uncertainty. The results for the clear days of the last 3 years at Villeneuve d'Ascq are presented, as well as a comparison with Total Ozone Mapping Spectrometer data. In most of the cases, both values agree to within 5%, but some cases present larger relative differences