Ozone profiles are derived from back scattered radiance spectra in the ultraviolet (289--339 nm) measured by the Global Ozone Monitoring Experiment (GOME) using the optimal estimation technique. Tropospheric Column Ozone (TCO) is directly derived using the known tropopause to divide the stratosphere and troposphere. To optimize the retrieval and improve the fitting precision needed for tropospheric ozone, we perform extensive wavelength and radiometric calibrations and improve forward model inputs. The a priori influence of retrieved TCO is ~15% in the tropics and increases to ~50% at high latitudes. The dominant error terms are the smoothing errors, instrumental random-noise errors, and systematic temperature errors. We compare our GOME retrievals with Earth-Probe Total Ozone Mapping Spectrometer (TOMS) Total column Ozone (TO), Dobson/Brewer (DB) TO, and ozonesonde TCO at 33 World Ozone and Ultraviolet Radiation Data Centre (WOUDC) stations between 71¿S and 75¿N during 1996--1999. The mean biases with TOMS and DB TO are within 6 DU (2%, 1 DU = 2.69 ¿ 1016 molecules cm-2) at most of the stations. The retrieved Tropospheric Column Ozone (TCO) captures most of the temporal variability in ozonesonde TCO; the mean biases are mostly within 3 DU (15%) and the standard deviations (1σ) are within 3--8 DU (13--27%). We also compare our retrieved ozone profiles above ~15 km against Stratospheric Aerosol and Gas Experiment II measurements from 1996 to 1999. The mean biases and standard deviations are usually within 15%.