Motivated by the desire to accurately derive tropospheric ozone from Total Ozone Mapping Spectrometer (TOMS) measurements, we investigate several aspects of these observations in the presence of highly reflecting clouds. Using the collocated Temperature Humidity InfraRed (THIR) measurements of cloud-top pressures, we identify three TOMS algorithm errors resulting from the inaccurate assignment of cloud-top pressure. The most significant error results from the inappropriate tropospheric ozone amount added below cloudy scenes to complete the total ozone column. After accounting for the cloud-height errors, we find significant total ozone column excesses of 10--15 Dobson units (DU) (1 DU=2.6867¿1016 molecules cm-2) over high-altitude, highly reflecting clouds compared with clear area observations. After accounting for additional algorithm errors involving the tropospheric ozone climatology and considering potential dynamical, photochemical, and NIMBUS-7/Earth Probe calibration errors, approximately 4-9 DU excesses over cloudy scenes remain. We speculate that the TOMS algorithm approximation of clouds as opaque Lambertian reflecting surfaces may account for a significant portion of these unexplained excesses. The excess ozone due to calibration error and unknown sources will affect the tropospheric ozone derived from TOMS measurements using clear/cloudy difference techniques. ¿ 2001 American Geophysical Union