A new algorithm to retrieve total column ozone from global spectral UV irradiance measurements is presented, and its accuracy is assessed. The expanded uncertainty (coverage factor 2) of the resulting ozone values varies between 2% and 3.5% for solar zenith angles (SZA) smaller than 75¿ and clear skies. For larger SZA the uncertainty becomes dominated by the sensitivity of the method to the atmospheric ozone distribution. Using this algorithm, ozone values were calculated from UV spectra measured by the National Science Foundation's SUV-100 spectroradiometer at Barrow, Alaska, between 1996 and 2001. Special attention was given to March--April 2001, the period when the campaign Total Ozone Measurements by Satellites, Sondes, and Spectrometers at Fairbanks (TOMS3F) took place. The data set was compared with observations by NASA's Earth Probe Total Ozone Mapping Spectrometer (TOMS) and a Dobson spectrophotometer operated by NOAA's Climate Monitoring and Diagnostics Laboratory (CMDL) at Barrow. On average, the new algorithm generates ozone values in spring 2.2% lower than TOMS observations and 1.8% higher than Dobson measurements. From the uncertainty budget and the comparison with TOMS and Dobson it can be concluded that ozone values retrieved from global UV spectra have a similar accuracy as observations with standard instrumentation used for ozone monitoring. The new data set can therefore be used for validation of other ozone data.