Abstract:We found that observed-data drifts are common phenomena by analyzing the multi-year temperature variation trends of 15 fluid observation wells in Qinghai province. The drift of 6 wells exceed 0.01 ℃/a, which is the limitation in the observed network. There are no correlations between temperature drift, water level and air-temperature, and no rule is found in drift distribution. We cannot draw conclusions about the relations between temperature drift, regional stress and strain status. The infection point of temperature trends are generated by the earthquakes or changes the observed positions in Yushu and Geermu wells, which implies that the observation system (instrument & sensor) is not the unique reason for temperature drift. To reject the drift by changing the means of observation, by comparing two different thermometers, is not possible.