The time series of mixing layer height and other characteristic variables in Naqu area are retrieved from Micro-Pulse Lidar (MPL, the same below) data by using the gradient method. Results show that the development of the mixing layer is slow in the morning and rapid in the afternoon, reaching a stable status at around 1400 BT (Beijing time). The fluctuation of the mixing layer height in the afternoon indicates that there exist frequent and strong convective activities over the Tibetan Plateau. An estimation of 400-500 m for the entrainment ozone depth and about 0.2 for the entrainment rate can be obtained by a parameterization method. Based on radiosonde data collected at 0800 BT and the daily highest surface potential temperature, the maximum mixing height (MMH) at 7 stations in the Tibetan Plateau can be obtained using the parcel method. The MMHs from MPL data and from radiosonde data show a good agreement with each other with the correlation coefficient of 0.85, mean bias of 0.11 km, root-mean-square error of 0.30 km, and the correlation passes the t test at the significance level of 0.05. The MMHs at the 7 stations all show an obvious daily variation. The annual mean of MMH indicates that there is no significant correlation between the MMH and the site altitude. The MMHs at Golmud and Dulan stations show different seasonal characteristics from those at other 5 stations. The former reaches their peaks in spring and monotonically decrease to low values in winter, while the latter reach their highest values in spring and the lowest values occur in summer. The above results show that terrain features of basin and mountain in the Tibetan Plateau have significant but different effects on the mixing layer height. With the definitions of thermal stability and characteristic pressure level, a statistical approach that can be used to easily obtain MMH is proposed based on the good linear relationship between thermal stability and MMH.