Tropospheric O<sub>3</sub> column estimates are produced and evaluated from spaceborne O<sub>3</sub> observations by the subtraction of assimilated O<sub>3</sub> profile observations from total column observations, the so-called Tropospheric O<sub>3</sub> ReAnalysis or TORA method. Here we apply the TORA method to six years (1996–2001) of ERS-2 GOME/TOMS total O<sub>3</sub> and ERS-2 GOME O<sub>3</sub> profile observations using the TM5 global chemistry-transport model with a linearized O<sub>3</sub> photochemistry parameterization scheme. <br><br> Free running TM5 simulations show good agreement with O<sub>3</sub> sonde observations in the upper-tropospheric and lower stratospheric region (UTLS), both for short day-to-day variability as well as for monthly means. The assimilation of GOME O<sub>3</sub> profile observations counteracts the mid-latitude stratospheric O<sub>3</sub> drift caused by the overstrong stratospheric meridional circulation in TM5. Assimilation of GOME O<sub>3</sub> profile observations also improves the bias and correlations in the tropical UTLS region but slightly degrades the model-to-sonde correlations and bias of extra-tropical UTLS. We suggest that this degradation is related to the large ground pixel size of the GOME O<sub>3</sub> measurements (960×100 km) in combination with retrieval and calibration errors. The added value of the assimilation of GOME O<sub>3</sub> profiles compared to stand-alone model simulations lays in the long term variations of stratospheric O<sub>3</sub>, not in short term synoptic variations. <br><br> The evaluation of daily and monthly tropospheric O<sub>3</sub> columns obtained from total column observations and using the TORA methodology shows that the use of GOME UV-VIS nadir O<sub>3</sub> profiles in combination with the spatial resolution of the model does not result in satisfactory residual tropospheric ozone columns.