Abstract: Long-term variations of total ozone at 4 proxy sites in China, Xianghe (XH), Kunming (KM), Mt. Waliguan (WLG) and Longfengshan (LFS), are investigated with ground- and space-based observations. Although there is an evident decreasing trend of total ozone since 1979 over the sites, the differences of total ozone between satellite and ground-based observations are distinguishable. The effect of cloud on the precision of total ozone from satellite measurements is analyzed with the reference of corresponding ground-based total ozone. The absolute errors of satellite total ozone data sets definitely increase in cloudy condition. Relative errors of satellite total ozone also exhibit their obvious increasing trends with growth of ground-scene cloud fractions, particularly at WLG, however the maximum of all the errors are less than 2%. Relative errors of TOMS total ozone vs. ground-scene cloud fraction show similar variation trend at XH and LFS where the sites are located in the typical middle latitude regions with climatology of high total column ozone, and the similar trend is also detected at KM and WLG where the sites are located in typical low latitude mountain regions with climatology of relatively low total column ozone. The relative errors of GOME total ozone vs. ground-scene cloud fraction do not indicate any regional attributes. The relative errors of GOME total ozone vs. FRESCO cloud fraction and cloud top height suggests that there is no obvious trend in the errors of GOME total ozone, and the errors increase only when the FRESCO cloud fraction is more than 50%. With the growth of FRESCO cloud top height, the relative errors of GOME total ozone exhibit an obvious increasing trend at XH and WLG with maximum around 3%. With the growth of TOMS effective reflectivity, the relative errors of TOMS total ozone increase with the maximum more than 2%. Generally, the TOMS-N7 total ozone over China is 2.0%-3.0% higher than that of TOMS-EP. It is speculated that there is a high possibility that the contribution of ozone within or below the cloud to the satellite total ozone has been over-estimated in the retrieval algorithms of satellite total ozone.