Measurements of BrO in the atmosphere are now routinely made using UV absorption spectroscopy, but it has recently been noted that the retrieved BrO amounts are critically dependent on the wavelength dependence of the ozone absorption cross sections used in the retrieval procedure (ozone absorbs in the same wavelength region and must therefore be accounted for in the analysis). Although atmospheric BrO measurements are typically made at a spectral resolution of ~1 nm, it has been shown that an inaccuracy of just 0.02 nm in the wavelength calibration of the ozone cross section can result in large errors in the retrieved BrO amount. Ozone absorption cross sections calibrated to <10-3 nm are presented here in the wavelength region generally used for the retrieval of BrO from observations of scattered solar radiation (345--360 nm), and for a wide range of temperatures (223--292 K). This level of accuracy in the wavelength calibration was achieved using Fourier transform spectroscopy, to record ozone absorption spectra at a maximum instrument resolution of 1.0 cm-1. Comparisons of these well-calibrated cross sections with those currently in general use for measurements of BrO, that is, those measured for the Global Ozone Monitoring Experiment (GOME) satellite instrument, reveal that the GOME cross sections must be shifted by an amount in the range +0.02 to +0.03 nm in order to correct their wavelength calibration. This correction to the wavelength calibration is shown to result in a significant reduction (10--20%) in the column amounts of BrO retrieved from ground-based measurements of zenith-scattered sunlight. An additional decrease of 5--10% is seen when the high-resolution cross sections reported here replace the GOME wavelength-corrected cross sections in the analysis procedure, reducing estimates of total atmospheric BrO correspondingly. ¿ 2001 American Geophysical Union