Global distributions of the tropospheric ozone column (TOC) and its error are estimated by assimilating satellite observations of total ozone column and stratospheric ozone profiles in a global chemistry-transport model. A quantitative comparison of the calculated TOC with collocated ozonesondes observations shows good agreement over a wide range of latitudes for the period studied (July--September 1992), even at the daily timescale. In most cases, the difference with the ozonesonde-based TOC is within the calculated error estimate. Comparison with the TOC distribution by Thompson and Hudson <1999> does not lead to any quantitative conclusion due to a mix of similarities and differences, even in the geographical distribution of the TOC. A positive TOC bias with respect to ozonesondes is found in this study over Hilo and Irene, probably due to a misrepresentation of the transport of ozone in a region of strong TOC gradients. In September 1992 the LIDAR observations from the TRACE-A campaign are also assimilated. This is shown to provide a constraint on the TOC that significantly propagates across the Atlantic Ocean, indicating the importance of point measurements of ozone for constraining its large-scale distribution.