Abstract: Global Ozone Monotoring Experiment (GOME) with the second Earth Remote Sencing (ERS-2) on board launched in 1995. To evaluate the performance of GOME's ozone data in China, ozone sonde observations at three stations in Lhasa (1998--1999), Xining (1996), and Beijing (2002--2003) are used to vali date ozone profiles and tropospheric column ozone retrieved from Global Ozone Monitoring Experiment (GOME). Ozone detecting systems are electrochemical concentration cell (ECC) sonde at Lhasa and Xining and GPSO3 sonde at Beijing; GOME data are retrieved by Liu et al. A comparison dataset consists of 51 matching pairs are obtained by applying a baseline criteria (±6 h, ±3° longitude, ±1.5° latitude). A statistical analysis of the differences between coincident O₃ profiles obtained by GOME and those obtained by ozone sondes are conducted using the methodology suggested by von Clarmann. When comparing with measurements the ozone sondes vertical resolution are much better than satellite retrievals. Retrieval averaging kernels are applied to the high-resolution data so that these data are comparable. As the high-resolu tion sondes profiles do not cover the GOME retrieval altitude range, the high-resolution profile (ozone sonde only up to about 30 km) are augmented with monthly mean climatological profiles from TOMS V8. Statistical bias determination and precision validation show that in the lower and middle troposphere, the mean biases are significant within 5% at Lhasa and Xining and within 10% at Beijing. In the upper tropo-sphere and lower stratosphere, the mean biases are within 10% at Lbasa and Xining and within 20% at Beijing. In the middle and upper stratosphere, the mean biases are within 5% at all three locations. The larger bias in the troposphere and lower stratosphere at Beijing may result from a different type of ozone sonde and different time period (ozone profiles shows multiple peaks in Beijing area in the spring). The mean biases of the tropospheric column ozone are within 10 at all three sites, which are partly caused by different estimated tropopause heights between GOME and ozone sondes. The GOME monthly mean ozone concentration at 0—2.5 km correlates well with surface ozone measurements. basically capturing the termporal variations of surface ozone at Lhasa, Waliguan, and Linan. In conclusion, from lower troposphere to upper stratosphere, GOME data used here has strong ability to reflect ozone distribution and dynamic changes in China.