Lidar-based ozone data sets may be meteorologically biased since the measurements are performed in the absence of precipitation and fog and the vertical profiles are truncated at the cloud base. This problem was studied for the routine ozone lidar facility at Bilthoven, Netherlands (52¿07'N, 5¿12'E). We used the long-term data record of balloon soundings at Uccle, Belgium (50¿48'N,4¿21'E) between 1969 and 1992 to examine the average ozone concentration as a function of altitude and season at different amounts of cloud cover. Since Uccle and Bilthoven are only 160 km apart and to a large extent have the same climatology, the Uccle data are assumed to be representative of the atmospheric conditions at Bilthoven. The data set of ozone profiles was arranged into several subsets according to increasing amounts of cloud cover. When compared to the total set, the subset for clear sky conditions showed significantly lower ozone partial pressures (absolute differences down to -2.37 mPa and percentage differences down to -43.6%) around the tropopause in spring and early summer and higher values (up to +1.10 mPa and +39.2%) in the lower troposphere in summer. These phenomena are explained by differences in the average origin of air masses and photochemical ozone production. The results have led to the definition of the monitoring strategy for the tropospheric ozone lidar facility at Bilthoven such that a meteorological bias in the set of ozone profiles is avoided.